Trends and emerging behaviours can affect an entire industry. This applies to your brand or company if you’re in the industry that’s seeing this shift. New opportunities, drastic shifts, or even a stable, long-term change to the industry situation can be the result. Here are some insights we have gained from a retrospective look at industry trends:

• Be strategic in Content Marketing and document it
• Have a dedicated Content Marketing Team
• Measure the effectiveness of Content Marketing through ROI accurately
• Prioritise the audience’s needs over yours when it comes to Content Marketing
• Instagram was the platform with the highest ROI

It’s all about communities (and valuable content)

• 54% of surveyed technology companies are confident in investing more in creating communities.

Companies should try more than ever to build a community that has an interest beyond just products/services. Values, vision, mission, and goals are a powerful mechanism to keep people engaged and interested.

Consumers want to know more about your company. Today more than ever, buying a product is often a personal statement. Therefore, companies should do all they can to build a community that has an interest beyond just products/services.

What do I do about this?

To build a community, it’s important to figure out what your specific audience finds interesting. More specifically, ask yourself who the company wants to target and why. Community building is not about imposing your will on potential customers. Rather, it is a two-way game of positioning yourself (in terms of values) and recognising what is important to your target audience. People out there build the community, not the employees, the CEO, or the company itself. Content should be created to match the current customer journey. Content should add value to the community.

Class, not mass

• Experts predict that a mass of (low quality) content will be produced with AI and flood the market.

Generative AI is now accessible to everyone and can be used for content creation. Therefore, many will not check or adjust the content in detail, which can lead to an immense wave of unqualified and undifferentiated (or even factually incorrect) content.

What do I do about this?

Generate high value content. What does high value mean? Tailor it to your target audience and their needs. You could do good and talk about it, show your workflows, and engage in discussions with your audience. Help readers reach their goal and understand the reason for their search and pain points. Ask yourself what is the benefit of your content for each of your audiences regarding their goals. Use AI as a tool not as the solution. HubSpot mentioned that Generative AI can be very good for idea generation, text improvement or information retrieval. At the same time, be critical, check the content, and give it a personal touch.

The new use of Social Media

Social Selling

• McKinsey predicts that in 2025 5.2% of all e-commerce will be  made by Social Commerce.
• Hubspot anticipate that in 2023 more sales will be generated via social media platforms than via the company’s own website.

Social media is increasingly being used for sales transactions. Social Selling is a major selling point in China and even in America it is too big to ignore. Someone scrolls on Instagram and sees a great piece of clothing or is reminded of their recently broken cable and has the opportunity to buy the products directly on the platform.

What do I do about this?

To sell on social media, brands need to establish trust on these platforms. Users ask questions about security, legitimacy and quality. This raises questions such as: Can I return it? Is it really genuine? Is it of the same quality as on the brand website? At this early companies can develop structures to exploit this new sales source.

Customer Service

• About 20% of GenZ, Millennials or Gen X have sent a message on social media to contact customer service.

Social Media is an easily accessible source for consumers to contact a company. Therefore, it’s no wonder that customers are increasingly contacting companies via social media.

What do I do about this?

Companies should train their customer service department on this and work closely with the social media team. As part of the emerging trend of “social selling,” these inquiries may become more frequent. See it as a chance to gain trust and to build a long term relationship with a customer, if you are able to solve their problems fast and with patience.

Social SEO: Social media as search engines

• Social media are supplementing search engines. This is a prediction from Hubspot.

The reason lies in the change in search behaviour. Consumers use various sources of information, including YouTube, Instagram, Facebook, etc.

What do I do about this?

Evolve your information points to take advantage of this as a company and stand out. The integration of keywords/hashtags is essential and an easy and consistent username is convenient for consumers to remember. Make it as easy as possible to find your company throughout the whole internet. You can use these sources for research as well to be up to date.

Your opportunity

Trends can change drastically from year to year, but they will still affect the way your business will grow and change in the coming year, and so it’s vital that you pay attention to what’s happening around you. Whether you’re a small enterprise or a large corporation, trends can still bite into, or create profits.

If you’re struggling to make sense of how to use these trends, we’re always happy to help, so drop us a line and let’s do this!

The post Navigating industry shifts with trends in content marketing appeared first on Switch - Digital & Brand.

“AI is not just the most important technology of our lifetime, but probably the most important in any lifetime. It’s going to be a new world of technology that’s as exciting as all the other worlds put together.”

Marc Bienoff, Salesforce CEO

That’s not a statement you can take lightly, especially since it’s coming from someone who knows a thing or two about the future of business: Salesforce powers a significant number of the world’s leading businesses, and it has access to the cores of the businesses that use it, so Bienoff’s visibility is pretty unrivalled.

The problem with statements like these, coming from men like him, however, is that they can be seen as the kind of comment you read somewhere but can’t ever dream of implementing.

Big data comes to mind as a classic example. There was a time, around 5 years ago, when everyone was extolling the virtues of big data, but it was the kind of thing that you couldn’t really envisage using if you were a small to medium-sized business. There were very few applications for us mere mortals, and even if we could find ways to use big data, there was no hope for us ever affording an implementation.

The same can be said of many of the other big transformative technologies that have seen the light of day over the past decade or so. The marketing world did see an element of democratisation in the form of cheap and easy-to-build websites, social media and ecommerce, but even then, our access to the tools didn’t mean that we, as small business owners, could compete toe-to-toe with brands that can throw much more money and resources behind the problem.

But generative AI is different. It’s true that there are massive benefits for custom applications built by corporate giants, but you and I have access to some of the most powerful tools available, either for free or for a very modest fee.

It’s becoming easier and easier to transform your business using AI applications, and you don’t need a technical background or any knowledge of programming to reap the rewards, all you need is an open mind and a healthy dose of curiosity.

How can I use AI for my small business?

There are hundreds of applications you could look for, and by the time this article is published the specific tools I’m using at the moment might have changed or have been superseded, so I’ll talk in broad strokes, rather than in specific instructions. Nevertheless, if you find any of the following tasks interesting, I’m sure that you can find the tools to leverage the lessons you can glean from here.

Here are some of the roles that I’ve used AI for to improve my processes dramatically. These have all helped me in one of two ways – they’ve either allowed me to do my job better or to save copious amounts of time on tasks I already did. In some cases, “doing my job better” was quite dramatic. I’ve done things over the past few months that I thought I could only ever do if I could afford expensive consultants (or little luxuries that I can’t justify, like a PA).

I’ll keep this list very simple, listing tools and processes that anyone can benefit from, even if you’re a solopreneur. We can delve into other, more complex ways to integrate AI in businesses in future issues, however today I want to keep this at a level where anyone can take this article and implement all the processes with very little effort and little to no cost.

What is Generative AI?

Before proceeding any further, though, I have to assume that some readers haven’t heard of, or maybe have heard of but haven’t used, generative AI at all. In a nutshell, Generative AI tools are tools that can create content from prompts you give it. Text-based ones, like ChatGPT (the most popular text-based Generative AI), can work on your prompts and give you text-based results. The wonder of these tools is that you speak to them in human language, just like you’d chat on email or messenger. So, you could give it a simple task like “Please explain why the sea is blue” and it will give you the answer, in conversational format, in text.

Image-based generative AI tools do the same but can output images instead – so you can ask for a beach scene with palm trees on the side of the picture, and the tool will create that image for you from scratch.

You can change the prompts and you’ll get different results, so you could ask it to explain why the sea is blue as if it was talking to a five-year-old kid, and the explanation would be simplified dramatically. If you wanted a much more detailed answer you could ask it to explain as if it was explaining it in a scientific journal.

Note: I’ll list a number of tools at the bottom of the article, with some differences and benefits. Keep in mind that when I use a name of a specific tool I’m generally referring to any of its competitors, too (so you can usually use ChatGPT, Google Bard, Claude.ai, HuggingFace Chat or Bing Chat for the same tasks and get similar results).

Using AI as a Business Consultant

This might seem like a big ask, but it’s actually one of the most powerful uses that I’ve found of ChatGPT is to ask it to brainstorm business ideas. The tool itself can’t take decisions on your behalf, but if you ask it to debate the pros and cons of the idea as if it were a business consultant, you’ll get a good starting point. Take the content with a pinch of salt, naturally, but I’ve found that it’s given me great ideas, it’s highlighted points that I hadn’t considered and in some cases, it actually changed my mind about some subjects.

You can also give it specific problems and ask for solutions – “I have issues with client retention, what would you advise? Imagine you’re a business consultant when you answer the question.”

Using AI as a Content Planning Assistant

This is one of the first, and most obvious uses of text-based tools. Well, actually, people ask the tool to actually write out content from scratch for them, but I think I’m a bit too particular about my writing to trust a machine to write for me (yet).

But I find ChatGPT and its competitors to be excellent content planning tools. I ask it to tell me what a popular article would include if I was writing about a specific subject. I’ll also ask it for an article outline, for some in-depth explanations, for some anecdotes to help me understand and for some potential pitfalls to avoid when writing an article about a subject.

I’ve also asked it to write out drafts for me, and even though I never used any of them, they’ve been helpful in my understanding of what should come next.

ChatGPT is also a fantastic assistant for title ideas and as an assistant for repurposing content (you can copy and paste an article and ask it to suggest a few LinkedIn posts from it, for example.)

Using AI for QA & Feedback

Once you’re done writing whatever you’re writing, whether you did it manually or with the help of a machine, you can paste your final content into the tool and ask it to proofread it. It can help with basic stuff like finding grammatical mistakes, but the real power, I find, is in the more analytical feedback I get from the tools. Sometimes they spot inconsistencies in my writing or actual issues with the content itself – recently it told me that I was bouncing between concepts in one of my linkedIn posts, so ChatGPT helped me write a more focused post.

Using AI tools as a Business Analyst

This is a slightly more advanced use, because you need ChatGPT Plus (which costs €20+VAT / Month), but boy is it worth it. ChatGPT Plus gives you access to slightly more advanced tools – first off you get access to ChatGPT 4, which is a more advanced engine. You also get the possibility to use plugins with ChatGPT – and one of them is built by OpenAI themselves: Code Interpreter.

If you upload an excel sheet to the code interpreter it will take all the data and make sense of it. First it will ask you a few basic questions about the data to ensure that it’s understanding it correctly. Once it’s confident that it’s figured it out you can ask it questions about the data, looking for relationships between numbers that would take you hours, or years, of tinkering if you were to do this in spreadsheets.

And the beauty of it all is that, just like we mentioned earlier, you’re only communicating with it in human language, there’s no training needed. As long as you ask the right questions it will answer them, and even if you don’t have questions – it can suggest a few highlights about your data that you might have missed.

And yes, I know that this doesn’t replace using a proper business intelligence solution, but for some companies, a proper business intelligence solution seems like an investment too steep to justify.

Using AI as a Virtual Assistant

I love this. I love how you can be simultaneously lazy and more productive in one simple action. I’ve started recording all my calls (on Fathom). This gives me transcripts of the whole call and a simple summary of the call. I usually take the full transcript and upload it to Claude.ai (I find Claude much better than ChatGPT for this use-case). The conversation is analysed and Claude gives you a summary of all that was said, the main points, and who raised them, and any action points that were agreed to on the call.

This has been a game changer for me in more ways than one. Having access to recordings has always been an option, but I never found the energy (or time) to go through them all. This way I can see a 5 minute summary of an hour-long call.

It’s also been transformative (no pun intended) in that I can now be completely present in a call, I don’t have the distraction of note-taking. Now I just take notes when I think of connections or new approaches, so I still keep my notepad open during calls.

Using AI to learn Something New

One of the last use cases I’m listing is also one of the most powerful. I’ve found that ChatGPT is great at creating course outlines for you, and can usually expand enough about subjects to give you a solid base to start looking further.

Just ask it to develop a course for you, give it the number of days or hours that you have and then ask it to give you the skeleton of the course. You can even ask it to start developing lessons, but while it’s great for less technical answers, there’s a point at which you should be handed over to books or more advanced literature.

I’ve used it a few times, and I must admit that I was impressed both by the depth it went into and by how good it was at explaining a complicated subject in layman’s terms. This is really one of my favourite use cases, because it gives me the capability to get up to speed with the basics of a subject before heading into a call with a client.

And in the end…

“Generative AI’s impact on productivity could add trillions of dollars in value to the global economy—and the era is just beginning.”

McKinsey

It’s important to note that this is a tiny, tiny portion of workflows that I’m using myself with zero input from external assistance. I got to these stages without pulling in developers, without disrupting any other members of the team or creating any new processes.

We’re using AI in other areas, and AI is embedded in a few of the tools that we’re already using, but I wanted to keep this as simple as possible to make it easy to kick off the AI revolution inside your organisation.

Just go for it. 

Some of the tools that I find invaluable

Generative AI tools – Text

• ChatGPT
• Google Bard
• Bing Chat
• HuggingFace Chat
• Claude AI (via VPN)

Recording

• Fathom.ai
• Google Meet

Visual

• Dall-E
• MidJourney
• Google Bard
• ChatGPT Plus (with Code Interpreter enabled)

AI-powered Search

• Bing
• Perplexity.ai

The post Generative AI for Marketing in B2B appeared first on Switch - Digital & Brand.

New tools have come and gone and the way we think of design has, while zigging and zagging, remained consistent over millennia. From the first tools and temples to the cities designed by the Greeks and the Romans and the Mayans, we have endeavoured to keep a healthy balance between form and function to solve the problems that surround us.

The printing press did not change the way we think about reading but it did give reading to the masses. A single new ‘design’ gave way to widespread literacy, changing the world and arguably leading directly to the internet.

But the longer answer relates to how we engage with design. It deals with the way the new tools are heading our way and how these could result in a significant shift in the way we perceive our roles in the design process.

Design thinking is expanding in scope

Design is a term with an immense scope. From drug design to graphic design, from residential interiors to Mars rovers, and from movie sets to fashion design, designers the world over apply their knowledge and the principles of good design to coming up with the stuff that shapes our world. All of these areas of practice are already making use of AI assistance. We have been successful in training machines to take on the problems that are too big for us to solve and to leverage huge computational power into solving the issue for us.

This is not new thinking. When we ran out of the limitations of our arms and legs, we designed and built the machines that carved out the industrial revolution. We could build taller structures, make more stuff and do so quickly, travel faster and farther, and in general unshackle ourselves from the physical limitations of our species.

Now, with AI taking a shape that more of us can use and understand, we are taking the next logical step and reaching beyond the limitations of our minds.

AI has been around for a while

AI is not new and it has been in use for a relatively long time in tech terms. If you use banking, social media, air travel, even the power grids in more advanced countries, you have been making use of neural networks with machine learning ability. The most efficient and hard-working systems that silently make our lives easier and more convenient are running some form of artificial intelligence but we are not able to look under the hood and notice.

What’s changed is that we can now ‘talk to’ an AI and get it to do our bidding. You’ve probably heard of DALL-E – the first image generator that we could speak to using plain English and that would generate images in a matter of seconds. A few months later it was the time for ChatGPT to make waves.

AI tools are more accessible than ever

But there are many, many more examples of AI-powered tools that are available to us, most of which have the ability to perform a single function very well. The easiest to get to grips with are the tools that exist as a subset of tools we use every day.

Take all the new neural filters that are inside Photoshop, or the integration of ChatGPT within Notion and Bing, the way photography stock libraries are giving access to generative image AI services, and other examples of the ready availability of AI that’s already making our lives easier.

There are also the useful tools that are making our workflow quicker and easier, the apps that take the repetitive or the boring and just make it happen while we’re doing other things. Designs.ai, Adobe Sensei, UIzard, Sketch2code, and all the apps out there that put a machine learning system at your disposal to automate the drudgery are there to give our minds better things to do. One could argue that these tools should and will become more pervasive as we apply our minds to really creative work.

Shifting behaviours

From a design perspective, there are more creative tools out there that are changing the way we work and, sometimes, the process with which we reach creative choices. Take using Midjourney or ChatGPT as a sparring partner, as a tool with which you can share ideas and dump the contents of an absurd idea onto to see what comes out. This is a behavioural shift. We previously ‘edited’ these ideas and then shared them with our peers. We pitched writing prompts, shared early sketch work, showed prototype models, sent over advanced drafts of a research paper… usually when it got to a point where we were seeking affirmation of our thinking more than anything.

Of course, doing so with an open mind, and selecting the people who we consider like-minded, made the process an integral part of the creative journey. Most design adds onto the work of our peers or of those who came before us.

Machines don’t judge, don’t get tired…

In the case of using AI as a member of the creative team, there is no need for affirmation, there is no fear of exhaustion or rejection or judgement – one can simply dump their intentions on whatever tool is available and see what comes out at the other end. Many times, it’s rubbish. Sometimes it’s predictable. But there are those times when the AI misfires in a way that is genuinely inspiring and this is where the magic can happen. At its least useful, it shortens the delay between thought and execution. And when we see it at its best it is a beautiful synergy between human and machine that creates what we’re already calling art.

This has been seen across the world of design. The impossible architecture, fashion, urban contexts, graphic art, and photography that is being created by the sudden rise of the AI artist is already an inspiration. It sows fear in those afraid of change but, to those with a grasp of the way the world works, it is a source of inspiration and imagination.

Work with the tools, not against them

The best work out there is being created by those who have taken to the tools like a fish takes to the sea and they have learned how to wrangle these unpredictable, often infuriating tools into submission, coaxing beauty out of the right combination of seed images and text prompts.

Once again, this is not new. Given the same paintbrush used to paint the Mona Lisa, I will still create a terrible stick figure. The best design has never been about the tools. It has always been about the brains that put the tool to task.

This time around, it’s a more complex and almost literally a more cerebral set of tools we’re dealing with so the knee-jerk reactions will be stronger. But it doesn’t change the way we think about design. It changes the way we integrate with the tools. And the sooner we realise that it’s up to us to make our silicon-based progeny help us create our best work, the sooner we will get to sharing it with the world.

The post 2023 is the year of AI. Could it change the way we think of design? appeared first on Switch - Digital & Brand.

Mark Zuckerberg’s dream for the future of the internet. An all-in-one experience where you live and work and entertain yourself. The next big leap into modernity for brands looking to attract new consumers. A tool for people looking to forge their way into a brighter future where they could truly be anyone they wanted to be. 

The metaverse: a seamless, interactive, internet-brought-to-life experience championed by AR, VR, and XR. 

In March 2023, a handful of months after hiring countless workers to boost Metaverse production, Mark Zuckerberg announced two pieces of news. 

1. Meta would layoff over 13% of its employees. 
2. Meta is going to focus on AI. 

Naturally, the internet took it as a clear indication that the metaverse dream is entirely dead in the water, and Meta threw away $36BN of profit. 

The truth is a little bit more complicated than that. 

Is the Metaverse dead?

No. For the Metaverse to be dead, there needs to be a complete stop in all activities related to the metaverse. No more activations. No more brands working on virtual reality store fronts. No more money pumped into the metaverse itself. 

There’s a second aspect. For the metaverse as Mark Zuckerberg envisioned to be dead, it would have needed to have been active to start with. 

The metaverse as Mark Zuckerberg imagined – that of a fully immersive environment – was never brought to fruition. The environment that Mark Zuckerberg wanted to create, where you would put on a headset and hop down into a world where you could feel as though you were there in person, was hampered by technological setbacks and the pure, unmitigating factor that that kind of immersive experience is years away from being even minorly accessible. 

As the metaverse stands now, it’s an extension of the internet with interesting AR and VR opportunities, but it was never anywhere near what the original video promised. 

And brands are still using the metaverse that was created: the metaverse that is just another extension of online activations that we’ve had before. 

Will the metaverse ever be created?

There’s always the chance that the metaverse will exist.

But not any time soon. 

The metaverse isn’t just as simple as saying ‘the internet, in reality’. It’s a combination of cutting edge technology that we simply don’t have available to us right at this moment, and that we won’t have available until more time has passed. Similar to James Cameron taking ten years to film the sequel of a popular movie because he had to wait until better technology was available, the metaverse isn’t fully dead yet, but its sleep might last for a long time. 

Why did Mark Zuckerberg stop developing the metaverse? Is it because of AI?

The official reason Zuckerberg gave is this: Meta now has a team that is dedicated to producing generative AI tools. This announcement, and the relative silence on the Metaverse, has led to the internet concluding that the metaverse has been paused indefinitely. 

It isn’t the worst thing in the world. The metaverse can’t move forward until the issues of better technology are breached. As a result, some stalling is expected. 

But AI likely has nothing to do with what’s going to happen with the Metaverse. 

In our first blog, we said that, although Zuckerberg is the one who named the Metaverse, it’s a bigger entity than just Meta, and that still holds. Meta itself isn’t the sole company developing metaverse technology, and new improvements can happen any day. 

It’s unlikely. 

But people also said that ‘thinking’ AI was unlikely as well, and generative AI made an impact in 2021 that hasn’t slowed down yet.

What about all the technology developed for the Metaverse?

They’re probably not going anywhere. 

Neither is that the metaverse suggested: the understanding that we can communicate on a much deeper level online. 

The problem with the metaverse isn’t a case of technology that isn’t useful in other ways. It’s a case of technology not being developed quickly enough to keep the momentum and hype going, and of being a steep price of admission for brands with very little possibility of payoff. The advancements made in VR and AR aren’t wasted: they’re new opportunities for brands to play with. 

However, it might be a while before we see them become the norm. They’re still expensive technologies to use, with little guarantee of significantly higher payoff – deep concerns for any brand operating during a global recession. 

Is AI the next big thing?

Yes. At least for now. 

However, generative AI models might find themselves hampered sooner rather than later: the industry itself is inundated with negative stories about the use of AI, and public backlash about generative AI has been swift. Brands that have opted to work generative AI into their regular line-up have to be careful how they do so: too much of it taking over traditional artforms, and the complaints are swift. 

Add in the current controversies about where, and how, language models are trained, and the capabilities of AI might be limited to certain aspects of businesses rather than become a wholesale creative approach. 

However, AI is undeniably a part of the future of marketing. 

Whether it’s going to be used to generate reports or write short captions, to come up with campaign ideas or to design simple things, AI has a place in the world we’re building. 

But its place is likely going to be less influential than we’re seeing right now. And AI – for all its capacity to emulate and respond to queries in a human way – is still a machine. 

Can AI be better than what it is now?

AI will improve: that much is certain. 

How it will improve is difficult to say.

The initial image-generative AI had a problem with hands. A program was written in a handful of days to train AI to spot this weakness and to fix it; therefore, now, a generated image that includes hands is more accurate than it was before. The journey for AI is going to be full of these setbacks and quick fixes, and we’re going to see the interim stages of AI develop in real time. 

What needs to be kept in mind is the public backlash against AI models. The program mentioned above received a lot of criticism levied at the idea that it’s training a machine to create artworks that real artists can do far better. This might prove to be AI’s biggest stumbling block – no matter how good AI can get, it’s always going to be dogged by the idea that the human element of art resides in the human. 

However, there’s hope for AI as a tool: to generate reports, to generate ideas, to come up with a quick way of saying something simple. 

For things more complex than that, the jury is still out. 

Some companies are already implementing AI into their systems…

When the Metaverse first became active, a handful of brands immediately hopped online to create their virtual worlds. 

This is a similar situation. 

AI isn’t refined enough yet to take over the amount of jobs that people are predicting. Similarly, the Metaverse isn’t refined enough yet to solely become a brand’s only presence online. 

At this implementing AI into the way your business runs isn’t a bad idea. 

It’s just not going to look the same between now and the next update to the generative AI models you’re using. 

That’s part of the process. 

Is AI likely to end up like the Metaverse?

No. 

The metaverse might be in stasis for a very long time, but the opportunity to improve on it is still relevant. 

However, our capacity to create generative AI models and to train those models is a far more accessible level of possibility than creating a new dimension of interaction. AI has almost half a century of headstart and is far more advanced than the Metaverse.

There’s another aspect to consider: AI and the Metaverse are going to collide. Tech doesn’t exist in a vacuum. The progress that is made in one will impact the other – whether or not we plan for it, and the progress that is being made for AI is going to change what we can do with  the metaverse. 

What else should I know about AI?

AI is still growing. 

This blog is going to be out of date by next week. The advancements made in AI move so quickly that we’re always going to be a little bit wrong about where it’s going, and when it’s going to get there – so our advice is this: keep an open mind. 

There’s a way to use AI that’s respectful, useful, and beneficial to your company, without generating the kind of backlash that other companies will generate. 

But the time to do that might not be now. 

In the meantime, all you can do is wait, and keep an eye on things. 

And the metaverse? What should I know about that?

This is a personal note. 

I think the metaverse is likely to happen – but I think it’ll look a lot more like the early days of the internet. Do you remember cluttered webpages, full of easter eggs and secrets? 

The thing about developing massive, structure-changing things like the metaverse is that if it doesn’t stick to the timeline originally envisioned, its development will slow down – but it doesn’t mean that the iteration already available won’t be used. In the Metaverse, the option to to create niche little corners of the web where secrets abound is already on us, and in some cases, it’s an option that people have taken to heart. There’s hidden comedy clubs. There’s places where you can play games. There are opportunities for you to meet strangers from around the world, and establish a routine online that’s different to the way the internet functions now. 

If you separate the metaverse from the idea of using it as a brand playground, there’s a lot of opportunity to be creative. It just won’t be fully immersive in the way that we’ve already seen – but that’s not a detriment. 

As it stands, there are fewer and fewer places online where you can go online and just be. A lot of the spaces that we remember from our youth have either become a shadow of their former selves, or they’ve rebranded and therefore lost their way, or their rebrand appeals to no-one because they try to court too many audiences. 

The metaverse is uncharted territory. At this moment, it can be whatever we make it, and everything online tends to go back to its roots, and that is both its greatest asset and its biggest downfall. 

For me, I think the Metaverse will be a resetting of the idea of online communication. I think we’ll go back to the nineties way of creating places just for fun instead of profit. I think the Metaverse will be tantamount to discovering a new way of communicating between us. 

But that’s only my take. 

Here’s another. 

The metaverse feels like an inevitable trajectory for tech, even if the shape it takes has yet to be fully determined. So far, we see it as a separate space. We consider the metaverse to be this parallel existence that we need to wear VR headsets to participate in. We see it as a sandbox within which humans (and brands) can exist in a way that’s not bound by the limitations of our corporeal existence. That alone can be fun. If you’ve played role-playing games or world simulators (remember The Sims?), you can see how a VR version of this could be hugely enticing.

Then there’s the AR version, an augmented version of our reality that’s simply superimposed by means of tech. This is even closer than we think and, as hardware shrinks in size and simultaneously gets more powerful, feels well within reach.

But this is only thinking about the tech. Consider the tech that is thrown our way and that we simply refuse to adopt. 3D televisions were all the rage for a minute but we didn’t buy them. Then it was curved screens but we didn’t like those either. Just because there exists a device that does something doesn’t mean we will all adopt it and it takes critical mass for tech to make financial sense. With Apple now making bold moves and their history of category-defining hardware, this aspect is moving closer to a reality we can grasp.

So, while we can’t speculate on what the form we adopt will be, we can be quite certain it will be the user experience that determines the success (or failure) of the metaverse. 

The post Metaverse is dead, long live AI? appeared first on Switch - Digital & Brand.

AI is going to take our jobs. AI is a bubble. Artificial General Intelligence (AGI) is just around the corner. AGI will never happen.

In reality, there is a little truth in all of the statements being made about AI but neither can be completely accurate because, like every industry caught in a frantic rush of development, it is impossible to see where all the pieces will land.

The AI space is vastly unpredictable

Even when the pieces land, the landscape will shift. AI isn’t a destination. It is a journey that our species will forever be intertwined with. We didn’t stop inventing machines when the Industrial Revolution had blown its last puff of steam. We kept inventing and aggregating. It’s unlikely that the inventors of the steam engine would have predicted Boston Dynamics and their robotic dogs but we can trace the lineage from one to the other.

Perhaps even more pertinent is the nature of the industry and the complex web of moving parts that it encompasses. Medicine, finance, search, advertising, aviation, energy, blockchain, imaging… the list of activities impacted is multiplied by the number of technologies being developed by all the established tech companies, universities, startups, and even individuals to leave us with so many moving parts that we can’t really nail anything down.

To add another layer of unpredictability, we must not forget that we are dealing with an intelligence. It keeps surprising us and bewildering is as we develop it. The idea of a predictable or explainable AI continues to largely elude us. While we know what we ask of an AI and can see what results it gives us, the inner machinations of deep neural networks that we have developed is too complex for us to understand. Add to this the “oh shit!” notion that we are using AI to develop better AI and it should cause us to pause for a moment before deciding that we know what’s next.

If there is one certainty, it is that this industry will keep surprising us.

Let’s look at jobs. As horrible as it sounds, some jobs are almost immediately at risk. And these are the ones, quite predictably, that most commonly resemble the tasks that the newly emerging AI models perform so well. But first, we need to have a shared understanding of what AI we’re referring to.

So what’s an AI in this context?

Let’s for a moment qualify what we mean by an AI.

An intelligence can be described as any entity that has the ability to choose a better outcome from a possible list of outcomes. The more intelligent this entity is, the more outcomes it can choose from and the better equipped it is to pick the best possible one. Using this model, an ant is not as intelligent as a mouse, and a mouse is not as intelligent as a human.

Humans have a broad spectrum of intelligence types – the same person can ride a bicycle in a city, write a novel, peel an orange, and flirt with another human using subtle, non-verbal cues. And the same human can generalise intelligence very quickly. We only need to burn our fingers once to know to avoid all flames, no matter what the source of the fire is.

Artificial Intelligence and the neural networks they run on, at least the ones we have access to today, work differently. They take a huge amount of training to generalise knowledge. Tesla have billions of miles of training on their self-driving car models to approach a reasonably good driving AI while it takes a human a few hours of lessons to achieve the same ability.

And they have a narrow spectrum of intelligence. The AIs at our disposal are really, really good at one thing. They are far superior to any human at that task but that’s all they know.

Armed with this knowledge, we can make a few assumptions about what tasks we will be passing on to AI or at least working with AI to help us do our job much better.

AI will help us do our jobs (and relieve us of annoying ones)

Let’s take a day in the life of a content writer or a designer. The day is usually split into many different ‘jobs’ or tasks. There’s some time dedicated to research, some time dedicated to thinking and putting ideas together, and some time devoted to actually creating something new. But there are also parts of the day that are relatively repetitive – the kind of tasks that we do without really needing to think too hard.

Let’s say a beautiful piece of artwork has been created. Now it is time to resize it to fit a pretty standard set of assets.

Or a wonderfully crafted blog post has been written and now it’s time to summarise it into brief captions for a handful of different social media platforms.

Both of these can be offloaded from our mind and given to a machine that has learned how do one of these tasks and to do it quite well. It’s been trained, by assigning weights to successful outcomes, to perform in a way that we like best. It is equipped to learn our preferences and to keep working that way. So, both of these creatives can work side-by-side with an AI and spend the time they’ve saved just enjoying their lives.

Now if there is a person whose job it is to do just the task that’s repetitive and that we would rather give to a machine to do, then that person’s job can be replaced by an AI. This is not to say it will, but it is possible.

But let’s take a plumber. They come home and troubleshoot the most bizarre of faults that we, messy humans, create in our own homes. They use their experience, their tools, their hands, and their ability to take instant decisions on the most practical outcome to get the job done and restore order and harmony in our homes.

Carry this over across all jobs that require skill and we can see that these jobs are safe for now. It will take us a while to implement the broader kind of intelligence needed to do these jobs and to build the robots it takes to give this general intelligence a useful corporeal form. To be more precise, the tech we need to build this robot probably exists, scattered around industries and intellectual property licenses, but it would cost as much as the GDP of a small nation to put one unit together and that’s not good business. For now.

It’s the mid-level white-collar jobs that are at stake.

Wherever there are groups of people in a bullpen doing what is mostly repetitive work that can be passed onto an automated system with a narrow but deep intelligence, the likelihood is that the populations of these bullpens will be the first to start thinning. Whether this is a good thing or not is up to your personal ethical standpoint and the subject of a significantly deeper discussion than this brief blog can go into.

A good point of departure is to think of the fate of the ecosystem around horse-drawn cabs at the dawn of the popularity of the motor car and work your way up from there.

We’re all coders

For a long time – meaning for all the time that computer code existed and all the way until a few months ago – programming a computer to do your bidding was the reserve of the coder, the software engineer, the guys and girls who knew languages that were esoteric to the rest of us. Now, we can address some of the most powerful computers out there by just typing a sentence into a chat box. In essence, every time we ask a chat-enabled AI for something, we are writing a little single-serve app and asking it to do our bidding. The more creative we are, and the more complex the task, the more useful and productive is the result of our interaction.

This means that natural language is already the de facto programming language. And as the AIs available to us get even more powerful, the need to use the programming languages of old (they’re so last week) will dwindle into obscurity. If you’re a sub-par coder, you’d better watch out because there’s a grandma out there doing a better job than you are and all she had to learn was how to write in English.

Symbiotic relationships are desirable

When we think of symbiotic relationships we usually go to that bird that eats the insects on a crocodile’s back as an example. The bird is fed and the crocodile isn’t itchy so both live quite happily. It’s an old definition of symbiosis but for the sake of this article we can put up with eye-rolling of biologists.

The happy relationship we’re referring to is the one between carbon-based lifeforms (us) and silicon-based brains (Artificial Intelligence).

The abilities of the systems that are available to us are quite staggering. They far outstrip the ability of any one person because they have been trained on gigantic models, huge data sets that represent the collective effort of hundreds of thousands of humans. They run on GPUs (the computers that power AI) that just love parallel processing and do so at outrageous speed. This means that, unlike us, they think of multiple things at the same time and do so thousands of times every second.

The highest ideal is, therefore, a human-machine tandem. A carbon-silicon creature that thinks with the value system and ambitions of a human and does so at the speed of a machine, drawing on a data set that is much broader than any amount of swotting for exams can achieve.

Think back to the industrial revolution. A lot of back-breaking work done by humans was replaced by machines that didn’t suffer and that could be significantly stronger than any one human. One man can carry one brick for so many metres but a steam engine can carry many, many bricks for as long as you can feed it coal.

What you want to be in that situation is able to drive the train or perhaps to repair it.

It takes skill to get the most out of any device, and AI assistants are a perfect embodiment of this maxim. Consider a camera. The same camera in the hands of an experienced photographer and in the hands of a complete novice will not produce the same photos. Give me the full set of paintbrushes that Michelangelo painted the Sistine Chapel ceiling with and I will produce an unholy mess that no pope would bless.

Embrace the race

Leaps in technology follow a pattern that echoes the one that came before. We’ve seen figures that show how long it took for radio to reach 80% of households and how that time was much shorter for TV, and even shorter for mobile internet.

And there is a pattern in the narrative, too. Early adopters tout a new tech as a panacea, gilding the advance with an unrealistic set of expectations of what it will do. Laggards and those afraid of change are vocal in their objections, with a handful of versions of ‘this tech will never catch on’ or ‘this tech is evil and should not be allowed’. While both positions are heartfelt, neither is entirely true of almost any innovation.

Photography purists are bemoaning the fact that some artists won photography competitions and immediately after they did, they admitted they’d created the images using Midjourney. What they’re angry about is that they didn’t have the ability to discern the difference themselves. ‘AI art isn’t art’ is an echo of the reaction that artists who had honed their skill at oils on canvas had when photography was invented. They argued that ‘photography isn’t art’ because photographers hadn’t bothered learning how to use a paintbrush. If an inkjet print of a photo fetches a couple of millions at the better galleries, why would an NFT of AI-generated art not do the same?

Midjourney won’t be replacing Saul Leiter’s colour work but it can already create passable street work that, at a glance, is coherent and convincing.

It is up to every one of us to land on the side of this story that is committed to history. A sensible place is to embrace the race, to get acquainted with the immense potential that these emerging technologies have to offer while being acutely aware of their limitations. This allows us to stretch the tech to its maximum, harnessing its power to make our lives easier and the output of our work even better. It also keeps us from falling into the trap of unrealistic reliance on a tool that has its inherent limitations.

Once again, this is an echo of past wins and losses with new tech. The ones who said new tech will never catch on died like Kodak did. The ones that put far too much faith in new tech died like so many of the tech startups at the dawn of the internet. The ones that embraced change and pivoted are still doing what they set out to do. We don’t think of IBM as International Business Machines any more but that’s what they are and they remained true to purpose. From cash registers when those were relevant to desktop PCs and laptops and all the way to Watson, they weren’t married to a specific technology but considered their purpose as delivering the machines that ran the business of the day.

We speak about being true to purpose at every opportunity. We also speak about business agility until someone tapes our mouths shut. But they are two maxims that have historically led to robust, sustainable business models.

Opportunity and agility

Opportunity lies in what you are already doing for your customers. Agility means you can remain valid to them, no matter what changes happen around you. Having your purpose clearly in mind means that you can be valid for as long as it takes us humans to evolve into a different species and that will take a while.

This means that to stave off redundancy, we need to be absolutely sure of what we are giving our customers. Let’s take Amazon as an example. They offer low prices, a wide variety of goods, and rapid delivery. Three fundamental pillars that humans will always want. In the words of Jeff Bezos:

“I very frequently get the question: ‘What’s going to change in the next 10 years?’ And that is a very interesting question; it’s a very common one. I almost never get the question: ‘What’s not going to change in the next 10 years?’

And I submit to you that that second question is actually the more important of the two – because you can build a business strategy around the things that are stable in time. … [I]n our retail business, we know that customers want low prices, and I know that’s going to be true 10 years from now. They want fast delivery; they want vast selection.

It’s impossible to imagine a future 10 years from now where a customer comes up and says, ‘Jeff I love Amazon; I just wish the prices were a little higher,’ [or] ‘I love Amazon; I just wish you’d deliver a little more slowly.’ Impossible.

And so the effort we put into those things, spinning those things up, we know the energy we put into it today will still be paying off dividends for our customers 10 years from now.

When you have something that you know is true, even over the long term, you can afford to put a lot of energy into it.”

This could be one interesting approach about your business to keep in mind in the face of a sea of uncertainty: What’s going to be constant in your space over the next ten years? What fundamental human need do you serve? Therein lies your opportunity and it takes all the agility you can muster to dance to the changing landscape and keep your operation relevant.

What should we be doing about AI?

If there is something we’ve learned about tech it is to never say never. ‘Never’ stems from fear of change. It is a wholly unrealistic standpoint and this is more true today than ever. The money pouring into silicon that can think is staggering. The parties at play mean business – quite literally. The quicker we accept and move ahead at the right pace the better. You don’t want to be the company that said “We don’t need to be on Facebook because it will never catch on.” Neither do you want to be the investor who put all their eggs into MySpace. But you do want to watch this space with a healthy balance of caution and optimism. Caution about the unrealistic claims and optimism about what machine learning systems can bring to your business.

If you’re recruiting in areas that are repetitive, consider investing in tech instead because automation of anything we humans do more than once will come sooner than we think. This isn’t new – automation has been around for almost two centuries now – but now it takes a deeper meaning. It goes beyond a simple ‘if this then that’ and delves into repetitive decision-making within narrow or specialised fields, especially when this thinking involves drawing on an overwhelming dataset.

Whatever your area of practice, there are Machine Learning systems being developed or already deployed, albeit at varying stages of maturity. Seek them out. If they appear underdeveloped or the results are as yet sub-par don’t dismiss them. Use them for inspiration. Riff on them. Perhaps more importantly, learn about them while they’re still embryonic and allow your learning curve to naturally grow with the tech.

When those who have taken the option to wait for some unspecified time in the future to figure out where the dust will settle, you want to be miles ahead. You want to have found the best way for silicon brains to do your bidding. The prosthetic superpowers we were promised in sci-fi narratives since Jules Verne put pen to paper are finally within reach. We’d be fools to leave them to others.

All images were created using Midjourney.

Here’s a little exercise. If you’ve gone through the cycle of being amazed at what ChatGPT 3 (and now 4) can do and then realised that it’s writing is completely devoid of humanity, read a paragraph of this blog at random and try and spot what it is that makes you think it was written by a human and not by ChatGPT. Put a reminder in your calendar for one year from the day you read this. On that day next year, look up a handful of blogs and do the same. Odds are you won’t be able to make this distinction any more.

The post AI have no idea what’s coming, but it’s coming appeared first on Switch - Digital & Brand.

Artificial Intelligence has the tech world in a twist.

It has those who are not in the tech world but aware of what’s coming in a pickle.

We’ve acknowledged that it is already changing the way we think, the way we work, and the way our jobs are going to be affected within a relatively short time.

Most people you know have at least one browser tab with an AI assistant running – ChatGPT, Notion with its ChatGPT integration, some form of stable diffusion/ midjourney… and we’re not even touching the AI that’s been running our lives as the fundamental underpinnings of Amazon, Facebook, smart banking, and all the other platforms that have deep neural networks integrated.

Just for the sake of clarity, in this article we’re dealing with the AI assistants that we mortals have the ability to speak to. The ones that have what we refer to as possessing a natural language interface.

The AIs that grew to run our world until last year (and continue to do so) needed people with deep knowledge of Machine Learning (ML) to wrangle the neural networks into submission. If you use Facebook, for instance, you may or may not be aware that it depends on ML for news feed personalisation, image and video recognition, translation, spam detection, prediction analysis, and much more.

And as dystopian as it sounds, the idea of an explainable neural network is still being discussed but we’re nowhere near there. This means that while those who build deep neural networks know what they’re asking of the machine and can see what it spits out, they have a very tenuous grasp on what happens inside. But that is another story for another day.

The dawn of the AI assistant

Let’s get back to the AI systems that you and I can easily use.

One way to speak about what’s going on is by drawing parallels with the history of the internet. If you’re old enough to have been there where it started, it will be easy to follow. If you’re not old enough, you’ve got a good handle on the early days of the internet from memes. We’re all good here.

To begin with, everything you’re reading is close to being outdated. By the time the pixels have dried on the last few words of this blog, things have changed. We definitely won’t be making any wild or speculative predictions. We know how that goes.

“I think there is a world market for maybe five computers.”

Thomas Watson, president of IBM, 1943

“I think there will be a cyborg brain in every kitchen.”

Pulling comparisons from the early days of the internet and perhaps even later on in the history of this space is easy because it serves as a historical reference point, a nod to the fact that history tends to echo and reverberate. But history gets more expensive every time it repeats itself. The ones that lose out to this game will do so burning a huge ton of cash the likes of which were not being thrown around in the nineties.

Let’s take a look at the meteoric rise of OpenAI’s ChatGPT. It took days for the news to spread and for a million users to flock to the chatbot, lured by the possibility of a genuinely useful artificial assistant. Within a week we’d found out the staggering depth of possibility it brought with it – it could write code, poems, university assignments, songs, stories and it seemed like we were only scratching the surface. And this is early days. Let’s face it – it is unlikely that even OpenAI had predicted the success. If you’re launching a product, one of the things you do is give your project a friendly name and ChatGPT isn’t exactly that.

We were also very quick to find its flaws. Of course, a large language model has bias inherent to the data set used to train it (and the humans who trained it).

Many of the perceived flaws were user errors. Those who failed to understand that we were dealing with a very specific language assistant and treated it as though an AGI had magically landed were disappointed. Why did ChatGPT suck at maths? Well, it wasn’t trained to do maths so placing an unrealistic expectation on it is the problem of the user not the machine. (Sidenote: WolframAlpha is great at maths but requires plenty of prompt jiggery pokery to be useful so bashing the natural language ability of a GPT-like language model with the mathematical wizardry of WolframAlpha is a matter of time or at least it should be.)

By the time we were happily using the bot for a handful of tasks, a number of things were apparent:

1. The training set was staggering in scope but ChatGPT4 was around the corner and will have a model that’s larger by several orders of magnitude.
2. The information it had access to seemed to encompass a history of humanity but the information sources ended in 2021 and the model had no access to real-time data.
3. The command of the English language was superb, even if the standard tone of voice is rather dull, but the bot started to break as soon as you tried any other language.
4. The speed at which the assistant operated was staggering but it quickly buckled under the strain of hordes of users hammering a machine with inherently finite resources.

A tiny technical detour

In the interest of brevity and to stay on topic, we’ll keep the technical aspects of these constraints out this discussion. But it is worth taking a brief detour to see how it takes sustained development and integration of systems to get where we are today and, a little later on this piece, what the immediate future could have in store.

AI got here in leaps. The history of machine learning seems to be characterised by cycles in which one giant leap that is followed by a long ‘winter’. That winter ends when the next leap happens, and so on and so forth.

Sometimes it is a hardware leap – the unexpected use of the GPU as a powerful enough processor to handle the immense needs of ML systems. Other instances are that of an advancement in the algorithm or neural architecture used, and other times we have access to new training sets. Back when ML was in its early days there was no internet so it is easy to see how much easier it is today to gain access to massive data sets we can use to train ML systems.

The transformation that the translator made possible

The most recent leap was the use of a language translator to translate from English to English – from words we speak as part of our day-to-day lives into a code the machine can understand and then back to English that we can read and understand. Hardware has been progressing steadily, leaving the venerable Moore’s law in its wake as it did so and AI models gobble up every ounce of processing power you can throw at them so the story of AI and that of processing power seems to be one that will be inevitably intertwined.

The tussle in the tech world

Back to the dramatis personae, the players on the scene this week and the ones that are shaping up to join the fray over the coming months. We’ve mentioned OpenAI and its wonderful ChatGPT. It has had university professors scrambling to prohibit it, others accepting it with open arms, and put Google into a veritable panic mode. Google actually used the term Code Red and ran back to its founders in an attempt to bring all human brains on deck as the battle with the silicon mind unfolded. And they’ve even rushed an AI event into our calendar during which they’re expected to launch their version of a chatbot. It will be live on YouTube tonight and we’ll update this when we know more.

What Microsoft is up to

Meanwhile, back at the ranch, Microsoft snuck up from behind and integrated ChatGPT into Teams and into Bing. Much is happening in that sentence so let’s break it down.

Microsoft is putting a few billion dollars into OpenAI. Rather than buying OpenAI, MS invested in the company and gained access to a symbiotic relationship with the AI giant. Within weeks, they’d put the gargantuan power of the Large Language Model behind ChatGPT into Microsoft Teams.

This means that after every Teams meeting, participants have meeting notes, action items, and highlights. These are generated automatically – we’re not talking of a simple transcript here – we’re dealing with an assistant that took notes, understood the salient points, and distributed highlights and action items. During the meeting, Teams can show a simultaneous translation of what anyone is saying in almost any language. And while this was a pipedream six months ago, it is already available and included in Teams Premium for $7 a month.

How about Bing? What is Bing, you might ask. It is Microsoft’s search engine, their idea of a rival to Google search that is used by an estimated 4% of the internet. During an event that saw Microsoft and OpenAI CEOs on the same it was confirmed that Bing will be powered by ‘ChatGPT 3.5’ and will initially be available for a few users and on Windows desktop computers only and within Microsoft’s own browser. It stands to reason that one would want a test-bed before unleashing this kind of power to a wider audience. Once this makes it to mobile, there will be no turning back.

Let’s for a moment break down two behaviours. Let’s say you want to buy a car and know nothing about cars. You can run a Google search for cars you should buy, perhaps adding parameters like “small, economical city car”, and then wade through the first ten links on the page of results. Google has, by its magical set of algorithms, presented you with the top ten sites that are most likely to be useful to you and it’s your task to read through pages and pages of articles or watch countless videos hoping to form a useful opinion.

Which, let’s face it, is not a fair representation of our reality.

You have another option. There is someone in your immediate social circles who knows cars, loves cars, and occasionally won’t shut up about cars even after you’ve glazed over. But if you call this person you will save plenty of time reading about a subject you care nothing about in search of a product you happen to need. They’ll know what questions to ask, would have read all there is to know voluntarily, and will zoom in on the best couple of options for you within minutes.

This is what Bing should be like with ChatGPT integrated into the search. A chat-like interface that gives you the power of an advanced AI while being linked to the entirety of the internet for real-time information. Ask a question in natural language and get chatting. Bing should be able to ask you follow-up questions – just like your friend would – to learn more about what you’re after and give you a more useful reply to your question rather than a list of links.

Google has been working on this for a long time but they have been taking their time to make it to the market. They’ve acknowledged that search could be much more useful time and time again, creating BERT, MUM, and other translator-based engines but have been stalling until it’s perfect, ethical, unbiased… and possibly too late.

We’re not predicting the future but it does appear that this could reimagine or supplement the traditional Google-style search sooner rather than later. Those of you old enough to remember the early internet will recall Google turning the tables on those who dominated the search space. We don’t mention Lycos, Altavista, Yahoo, and Excite when speaking about search nowadays but they were the ones who set us up for a searchable internet and they did so by guiding us towards the most appropriate UI and the most useful search results. And Microsoft was clear about their intent:

“It’s a new day in search… The race starts today, and we’re going to move and move fast. Most importantly, we want to have a lot of fun innovating again in search, because it’s high time.”

Satya Nadella, Microsoft CEO, 2023.

Google is determined to stay in the game

Google is making all the right moves as well, with their AI announcement that speaks about them integrating their Language Model for Dialogue Applications (or LaMDA for short) along with a whole lot of other news. That announcement felt so much like Google is really feeling the burn of the openAI launch and media buzz. They’ve been hacking away at AI for more than a decade now, their models made openAI work (remember it was their transformer that unleashed the beast) but they took the cautious route, concerned about a perfect product and an ethical approach. Google has more up its sleeves than anyone, and yet OpenAI simply made the news.

There’s much more to AI than the tech giants though. Anthropic AI is one of the smaller players. Or, to be a little more up-to-the-minute, Anthropic AI was one of the smaller players. It’s claim, one that hasn’t been updated in light of recent events, remains the following:

“Anthropic is an AI safety and research company that’s working to build reliable, interpretable, and steerable AI systems. Large, general systems of today can have significant benefits, but can also be unpredictable, unreliable, and opaque: our goal is to make progress on these issues. For now, we’re primarily focused on research towards these goals; down the road, we foresee many opportunities for our work to create value commercially and for public benefit.”

But Google just dropped $300 million into Anthropic so those down the road opportunities seem to have come to fruition rather rapidly.

But there’s more to tech than the giants

LAION (Large-scale Artificial Intelligence Open Network) is also currently steaming ahead with its open-sourced AI assistant and is actively crowdsourcing training efforts so that anyone on the web can contribute to an always-free model. For those of you old enough to remember MS Encarta, this is like Wikipedia is already in advanced stages of development and population before Encarta had progressed from beta. LAION are working on several models, including image generation models, and they’re all open sourced.

By now you will have noticed that many of the players mentioned were unknown outside tech circles just a few months ago. And the pace at which the new players gain ground is quite staggering.

The image generators

Just look at the image generators. DALL-E, another project by OpenAI, was impressive enough when the beta landed half way through 2022. By the end of 2022, we’d not only seen a flurry of stable diffusion models that put it to shame but also the commercially successful launch of the clearly superior Midjourney. In the space of 6 months we were introduced to generative models and went all to way to an ecosystem that is developing around some pretty impressive technology. There are already some artists selling courses in AI image prompts.

And most of these are still in beta. If you thought there was model bias in language models, you should give image generation AIs a try and prepare for some cringe. So, while we’re on generative models that have been trained on a massive data set, it is early days and we’re all using the same data set.

Hey Midjourney, how about a portrait?

No matter how many times you try, “the most beautiful woman on earth” is always white.

There are already versions of Stable Diffusion that anyone can download and run on their machine – provided they have the hardware to do the heavy lifting. Consider that the latest Mac Mini to hit the shelves of consumer stores has a transistor count that exceeds 40 billion and costs around $1300. That includes a 16-core processor dedicated to machine learning processing. Speaking of Apple and their ML hardware we can’t miss the forest for the trees. So far, their hardware tackles relatively simple stuff like Siri, image signal processing for mobile camera systems, and other relatively low-demand applications for this gargantuan processing power. But Apple is something of a dark horse in this arena. All that power is almost doubtless to be waiting for something significant. Their first all-hands internal in-person announcement since pre-covid happens next week and it’s been dubbed the ‘WWDC for AI’. We’ll update this space once we have more details.

In any case, the massive processing power available to mere mortals means that models, or image sets that can be used to train and generate your own imagery, are about to be a ‘product’ that has unbelievable room for customisation. We haven’t scratched the surface here yet but proprietary models are as inevitable as the stock photo library is. (Or was, depending on when you’re reading this.)

As with all new tech, the early adopters and those willing to put in the hours to learn how to optimise their interface with the tech are those who achieve the best results. But this does not mean that it will be the inevitable trajectory.

Looking back to peek ahead

Once again, we look at the past to hope to understand what’s coming. The early internet was for the tinkerers. It was for those who knew how to make the damn thing work and what port the internet box thing plugged into. That didn’t last long. Great technology is initially indistinguishable from magic. Then it becomes even better when we stop regarding it as technology because it becomes easy and pervasive.

This step has already happened. Writing this into Notion means we can hit /Help me write and ask for help with practically anything. MS Teams, whatever Google is coming up with next, and the tech that the rest of 2023 will bring will be instantly usable to everyone because it took all of six months for a natural language interface with AI to go from tinkerer-only to general public.

AI has the tech world in a twist but the tech world has always been in some twist or another. There have always been Kodaks and Nokias and Myspaces and Palms. What we can’t say is if there will always be Apple and Google and Microsoft and Meta.

That is up to us. We, the users of tech, get to pick what we want and it isn’t always what’s the best from a technical perspective (BluRay has much better picture quality than Netflix) because we adopt what’s convenient, is in the right place (usually our phone), does the job, is priced well, and doesn’t need much technical knowledge to use it.

We can’t possibly predict what’s next but we’re watching the space attentively. And we’re doing so from a number of angles. From the lazy/optimist perspective, we’re glad to have machines that will help us do the work we do, that can act as technological prostheses that carry on from the limit of our mental capacity and extend it beyond where we thought possible.

From an ethical standpoint, we do hope that the tools that are headed our way will be equitable in that they will be available to all. We mentioned Wikipedia as a great example of this, a sign of the internet times that is a shining beacon of crowdsourced and publicly available knowledge. We look forward to seeing the next evolution of AI inhabiting a similar space.

There will be winners and there will be losers.

We the users will be winners as AI becomes more ubiquitous, more useful, and dare we say, more open. The losers will be the tech giants that don’t play their cards right. Unfortunately for them, this game is one where the rules are still being written, where the dice have a constantly changing number of sides, and to which players can simply join unannounced and mid-game.

We just can’t wait to see how it all unfolds.

The post AI: there will be winners and losers appeared first on Switch - Digital & Brand.

They won’t attempt a conversation with me about football that was played over the weekend, knowing I’ll have no clue – I don’t watch football. So, because they’re good friends, they don’t ask me to join when they’re watching a match. They realise that in the Venn diagram of our respective interests, football does not intersect. 

This applies to most human relationships. We interact and, while doing so, share information about ourselves. The type of coffee I order shares my preference data with the barista and that allows for them to prepare the coffee I like. That’s as much as I share though. Anything else would not necessarily intersect. 

We share much more data with close friends, family members, and significant others. Yet there is no one person who knows absolutely everything about us.

Enter the AIs that rule my world. Google, Amazon, eBay, Facebook – the lot – are different. They know more about me than any of my closest friends or family. After all, in the 21st century, humans exist to convert experiences into data that can be shared with other humans.

Most of our day is digitised to an extent we’re not actively aware of. Take a photo with your smartphone. It is analysed to determine its content and geotagged. If you search your photos for ‘dog’, ‘seaside’ or even “shellfish” your phone will give you a pretty accurate list of photos you’ve taken that it was busy analysing and categorising for your eventual convenience.

The same applies to the thousands of words we type into our connected devices all day – our chats, searches, social media interactions, emails – but it goes deeper. Our swipes and scrolls and the amount of time our cursor hovers over a specific part of a website is all ferreted away to contribute to the algorithms that are exceptionally skilful at gaining a deeper awareness of our thought processes.

So, while we share limited amounts of information with every human we interact with, our digital profile is probably the most complete assessment of our current state along with a comprehensive history of our past behaviours.

This might sound like a warning. It is not. There is no single massive AI that has access to absolutely everything and, more importantly, there is no algorithm (yet) that has an intent other than to use our data to improve our user experience and to tailor the targeting of promotional messages.

Which brings us to the flipside of the knowledge that AI has of our state. We can use it to be significantly more effective, and efficient, at creating and distributing our brand’s messages. With the mind-bogglingly detailed and granular information that’s available about every internet user, the blanket-bomb approach to marketing and advertising suddenly appears terribly archaic.

As marketing practitioners, we find ourselves in an awkward conundrum. Should we tailor our communication to humans or to the AI, such as Facebook and Google algorithms, that ultimately serve our messages to the humans who use the platforms? The answer lies in a carefully crafted strategy that does a bit of both. And it goes further – we listen to what humans are saying by analysing their behaviour to better serve the AI that acts as a delivery system.

This creates a virtuous cycle by which we analyse the performance of campaigns that are running, a real-time listening process, and further improve our messaging as we go along. In our white paper about Performance Marketing we go into the technicalities of this approach. If you spend 10c on digital marketing (or more than 10c) I strongly suggest you read this because it can help you dramatically increase the return on your marketing investment.

As we progress towards a more targeted form of marketing we approach a state that was portrayed effectively in the 2002 movie Minority Report. What was then sci-fi showed the protagonist walk through a shopping mall in which all advertising was tuned to him specifically, addressing him by name and suggesting products that would be interesting to him – including tank tops at Gap that would look great on a rather buff Tom Cruise. Two decades ago this was considered a wildly improbable scenario but the holy grail, the audience-of-one, is fast approaching. And as it does, your brand needs to be well placed to cut through the noise and deliver its messages in a way that’s specific, respectful, and completely relevant to its audience.

We hear brands complain that there is too much noise, that audiences are bombarded with so much promotional content that they tune it out. Done properly, we could very well reach a stage where our audiences finally feel that marketing is no longer tone-deaf, that it has finally understood the needs of the human and delivered just what we wanted when we wanted it. Maybe AI is the friend we truly need after all.

The post My friends and AI; Or will AI be the marketing tool that we finally make friends with? appeared first on Switch - Digital & Brand.