With the worldwide craze surrounding Generative AI, it’s hard to imagine that this actually only started a year ago. And it won’t go away anytime soon. The number of new cool tools and upgrades to existing ones in the past twelve months has been astonishing. Every time you think you know all the good tools and functionalities, but at the end of the day that knowledge is already outdated. In the coming year it seems that this speed of developments will not slow down. On the contrary. Writing a trend blog for 2024 is an interesting task, but I’m going to try it!

Sixteen years ago, when the iPhone was introduced, did you expect that a decade later we would be staring at it for hours a day? Navigate, edit photos and videos, bank, video calling. The cool thing about many technologies is that we often don’t even know or could predict the coolest functions yet. You now see that with Generative AI, but you also saw that with crypto, for example; that was introduced as purely a peer-to-peer cash system. The next generation of the internet is now being built with it and there are numerous offshoots such as NFTs, CBDCs and DeFi.

The superlatives surrounding GenAI also continue to fall short of the predictions of major consultants and banks. Because it is ‘the next productivity frontier’ according to McKinsey, this could save companies $2.6-$4.4 trillion next year. According to Goldman Sachs, this will also result in an increase in the value of goods and services of 7%. Research from KPMG also clearly shows this; 73% of companies expect GenAI to increase employee productivity in the coming year and 71% want to implement the technology in the coming year.

As the motto of one of the best-known startup incubators Y-combinator puts it so beautifully; “Make Something People Want”. When I talk about GenAI, I always see plenty of hands raised from people who have not yet played with a GenAI tool or even work with it on a daily basis. But this is not because they don’t want to. Every time I ask questions, the main thing I hear is that developments are going too fast for many people and they don’t know where to start. It’s like going to a tribe of Indians somewhere in the jungle and pushing a smartphone into people’s hands. But with the multitude of courses and most importantly; the ease with which you can start working immediately and free of charge will cause adoption figures to rise unprecedentedly in the coming year.

The renaissance of our work and workplace

In my opinion, AI will really fundamentally change the human-machine relationship in the workplace. While the focus this past year has been on how GenAI can help with content creation, I expect to see much broader adoption next year. Not just generating beautiful newsletters and images, but really helping with advanced analytics, predictive insights for projects and business management, software development, greatly improved customer service and facilitating more effective outcomes such as in healthcare.

Make better and faster decisions, because you no longer have to wait for the opinion of a colleague or supplier. You no longer have to think about whether it is worth it to outsource a question, because you will not immediately receive an invoice for the advice. You can now make analyzes based on your own data, which you could never make yourself before.

I now have customers who include GenAI as standard in meetings and brainstorms and, because of the results, are even considering minimizing the number of planned brainstorms and leaving this to the technology. I have friends who love that they get so much good code at the start of the day by playing with CoPilot for half an hour, that they can do their own things for the rest of the day without the boss noticing.

There are of course plenty of other examples of the impact on work; from companies that have replaced entire departments of staff with GenAI to companies that have uploaded their entire data archive into GenAI, which now serves as a basic knowledge source for employees. McKinsey launched Lilli, JP Morgan JPM and uploaded hundreds of thousands of research reports and other useful PDF files. Employees worldwide can now log in and ask the chatbot for advice.

But this rapid progress, in my view, makes the redefinition of work and functions necessary in the coming year. Because instead of getting completely sucked into the GenAI flow and working with the tools all day long, I think it is important as a professional to first take a step back and recalibrate in 2024.

Which activities would you prefer to delegate, eliminate or outsource? Which activities secretly cost you a lot of time? What often contains the most slack in a project? What analyzes would you like to make, with all the data and documents you have? Which activities that I cannot perform myself are important for achieving my goals next year?

FOBO sharpens the mind!

With these first questions, I took a close look at my own companies and activities and selected 30 tools via ThereIsAnAIforThat to try out, which ultimately left me with 8 that I use every week and 2 every month. Not only for generating and optimizing content, but also continuous analyses, brainstorming and enriching my product offering to customers.

In my search for good tools I have had FOBO (fear of being obsolete) a few times; I saw AI tools that gave such great results in seconds that they simply overshadowed some of the consulting activities that I marketed myself. I now simply offer these tools proactively to customers, with of course guidance on how to use them properly, in a broader package of services. The customers are even happier and my work has not deteriorated. Don’t be afraid that AI will also take over your own activities, but really worry about how you can make other activities even much stronger.

Back to the prompt table

As humans we are simply lazy; just look at the most searched search terms on Google (‘how can I do the fastest X, easiest X, etc.). According to research, our brains are programmed in this way so that we use as little energy as possible. This does not only have to do with people who do not exercise as a result. I work out every morning, but in many areas I am looking for anything but a goat path. According to scientists, we call many such activities ‘efficient’, but in principle we are just exhibiting lazy behavior.

I see the same thing with many professionals who work with prompts. Why make things difficult when you can just copy a prompt from a blog or use one from a colleague, which will give you a result you’re happy with? I often receive feedback from customers that they are dissatisfied with the output of various GenAI tools. But that’s usually because the prompts are too general or don’t fit the assignment.

We have been able to experiment with the many tools over the past year, in my view 2024 is the opposite of a Prompt Paradigm Shift. Professionals who want to get ahead of the pack are going back to the drawing board and scrutinizing their prompts very closely. Looking at all possible factors that can still be added or tightened up. For example, I now use a prompt of ¾ A4 for my SEO work. In addition, I no longer rely on one prompt that will do the trick right away, but rather a series of prompts that lead to the perfect result. We see a wonderful example of this at NASA.

Beware of BYOAI colleagues who work in the shadows

The GPT convenience serves working people incredibly well. But many employees forget that much of the data they use is confidential. Research shows that 11% of the data that people paste into GPT is actually confidential data. Which can cause very unpleasant effects. For example, Samsung discovered that developers were using GPT to test code, because GPT suddenly came up with highly confidential source code from the tech company. According to research by Dell, 91% of professionals have already worked with GenAI. Just think about how much confidential data is already stored in the bot.

There are plenty of organizations that have already blocked GPT for employees, such as Dutch ministries, banks and countless companies. Not only for fear of leaking confidential data, but also because of, for example, lack of clarity about privacy and regulatory issues. But I hear plenty of customers who still use the tools, because the unprecedented power not only saves a lot of time, but also really enriches the work itself in terms of quality. From hotel chains to supply companies. Real ‘bring-your-own-AI’ (BYOAI), without this actually being allowed by employers.

Here I see an undesirable development emerging; the formation of Shadow AI. We previously saw Shadow IT, where individuals or even entire departments within a company saw the power of certain IT and started using it without authorization. With all the associated safety risks. This happened a lot, for example, with cloud data and Software-As-A-Service.

Man is lazy, but not stupid. I was recently working with a customer in his office and we wanted to look at an AI tool together. This was blocked on the network and therefore could not be accessed on the phone via WiFi. Indeed; two seconds later the WiFi was disabled and the tool was used on the 5G network without any problems. Are you going to stop something like that? Impossible. It does indicate the urgency of a good protocol that you record with all colleagues.

Multimodal will become the new normal

AI has been used in large organizations, such as healthcare, for many years. With the healthcare crisis we are heading for in the Netherlands, smart tools that save doctors’ time and also allow them to work more efficiently are very welcome. A good example of an AI that has a major impact on this is Google’s breast cancer detection. A process from initial detection to results, which can easily take two weeks. But with Google’s tool you get a result within a minute, with an accuracy of 99.98%.

This is an example of a ‘multi modal’ AI; a tool that can also see, hear and talk. Until now, most GenAI tools specialized in simulating a single form of expression. However, with the arrival of models such as GPT-4, the trend in GenAI is towards multimodality. For example, Meta has already shown a GenAI model (ImageBind) that can simultaneously bring together images, text, audio and temperature in an analysis.

These tools are also already widely used; I recently spoke to a pharmaceutical company, where most doctors in the room admitted that they sometimes throw an MRI into GPT for a second opinion. But I also see owners of web shops, for example, working hard on this. For example, by uploading a photo of a product and getting a perfect SEO-optimized description for the online store. I personally use tools such as Otter.ai for taking minutes of appointments. Glif is also worth a try, which allows you to generate a whole set of art, cartoon drawings, selfies and more with a simple prompt.

In addition, I see funny examples every day of things that would never have been possible before, but suddenly become possible with multimodal AI:

Of course, this can also be used on a much more advanced and large scale. In my opinion, these new models open doors to unprecedented new possibilities with the use of GenAI.

Google has already created a multimodal system, which can predict the next dialogue in a video clip. Car manufacturers use it for the ongoing situation analysis of self-driving cars. Nine out of ten Dutch hospitals use it, mainly to determine the best treatment.

It’s time to act with an act!

Brussels is once again realizing the ‘Brussels effect’; it is the first major government worldwide to introduce far-reaching legislation and regulations. The AI ​​Act classifies the various applications per risk, which in turn is linked to the strictness of the rules. In addition, the tools must also comply with European copyright and European transparency obligations.

I once again read all kinds of ‘breaking’ messages on Linkedin from people who think that this completely limits innovation within the EU. But just like with the MICAR legislation, the European legislation and regulations in the field of cryptocurrency, which will come in handy next year, I see many advantages. Just as with cryptocurrencies, I hear from plenty of customers that they are waiting to use the technology until there is more clarity in terms of legislation and regulations. So that will be next year.

There are simply enough bad actors in the technology industry. In my view, well-thought-out legislation will ensure that the benefits of AI are fully realized while protecting the fundamental rights and safety of users. In addition, companies that use the technology should also take a good look at possible risks.

I am very bad at remembering names and had hoped that I could pair the smart Rayban-Meta glasses with Pimeyes. Unfortunately; matters such as real-time identification of people based on biometric data are not permitted under this legislation. Probably to avoid Chinese situations.

Together they are even stronger

In my view, the big next development in tech is convergence; the coming together of technologies that ensure an even more powerful symbiosis. That is not only nice, but also very necessary. Just look at the early years of cryptocurrency; According to research, the underlying blockchain technology used as much energy worldwide as the total energy consumption of the Netherlands. A problem that has already been largely solved with ‘the merge’ (the most used blockchain Ethereum reduced its energy consumption by 99.8% due to a major upgrade, which I wrote about earlier).

The American MIT recently released a study showing that generating one image with Midjourney costs the same amount of electricity as fully charging the phone. Google recently indicated that GenAI accounts for 10-15% of its global energy consumption. Microsoft also saw this energy use with its own AI tools and is therefore now even looking at its own nuclear power plants to provide the required energy consumption in the coming years.

Fortunately, in the field of GenAI, we are now seeing great convergence with ‘edge computing’, which brings the data storage and the IT tool closer to the user. So much more decentralized, instead of central.

Blockchain is also seen as a great friend of GenAI. Blockchain ensures secure storage of data, which can be used in many different ways. It can help validate the authenticity of images, videos, documents and other types of media by being able to cryptographically verify where a piece of content came from and whether it has been tampered or altered in any way. This is done through so-called Timestamping, which was recently also used to prevent fraud during the elections in Guatemala.

The many GenAI tools rely on large data sets. There are plenty of people who wonder; Why would I give my data to such a tool if I am not rewarded for it? This is where a beautiful convergence emerges between GenAI and cryptocurrencies. Germany’s Bosch rewards users for sharing data with cryptocurrency. Something that is made possible by Fetch.ai, whose CEO and founder recently spoke to me at an event. He was one of the founders of Deepmind (which was acquired by Google and transformed into the AI ​​department of the tech giant) and is now building solutions with Fetch to easily make money by sharing data.

Next year, quantum technology will also make AI much more powerful.

Quantum technology provides exponentially greater computing power for computers. This allows AI to process, work and analyze data much faster. We are not talking about simply finding a break in an MRI scan, but really testing new medicines and managing financial markets through simulations. GenAI on steroids. More and more platforms are being added that you as an organization can use immediately, such as Sandbox and Xanadu.

And young people? Virtual world builders, such as the very popular Roblox, have already made countless GenAI tools available and will greatly expand this next year.

Don’t forget ethics!

With all the great and cool tools, we still too often forget the ethical aspects. The EU’s AI Act already provides more transparency and that is really important. In my view, it will quickly become more difficult to ignore concerns about its impact on privacy, data protection and business operations in the coming year. Especially about how to handle GenAI responsibly.

Here too there is convergence; the ethical issue is a responsibility that must be shared by everyone: developers who design and market the GenAI models, users who use them for different applications and legislators who determine the rules of the game. Here, a much better balance really needs to be found between maximizing the benefits of generative AI and protecting our core human values.

That is quite a task, as has been evident over the past year. Numerous content creators, from well-known book authors to photo stocks, have sued the makers of GenAI tools, because they believe that their content should not simply be used for training AI. Something that the makers of the tools do not agree with. But also the ‘hallicunations’, which can sometimes provide hilarious but also increasingly shocking, offensive content. Many policymakers are already squinting at next year, when half of the world’s population will be allowed to go to the polls. What impact will GenAI have on this? The possible consequences of incorrect use raise numerous ethical issues.

Now the biggest challenges on the web are often also very big business opportunities. Tools such as those from Delft-based DuckDuckGo offer solutions for this, large companies are setting up their own ‘AI ethical board’, global Ethical-AI consortia are being launched and the EU AI act provides a number of additional tools. Next year will be an important test in that regard.

Making it completely ethical will never work. You also see this with countless other technologies. Legislation and regulations surrounding ethics remain a cat-and-mouse game between the ‘bad actors’ and governments. As a speaker at an event I recently said so beautifully; ‘it’s a journey, not a destination; a continuous process of developing, learning and adapting’.

With the speed at which developments in the field of GenAI are currently taking place, I am curious to see what the predictions will become and, above all; what unexpected but impactful developments we will see unfold in 2024. As Peter Drucker so beautifully puts it, “the unexpected was the richest source of opportunity for successful innovation. Unfortunately, expected occurrences are not only many times neglected, but are frequently actively rejected so they are never exploited as innovations.”