GPT is presented to businesses on a silver platter, immediately available for use,” says Ondřej Vaněk, Chief AI Officer, Adastra

Ondřej Vaněk, Global Chief AI Officer of Adastra and CEO of Blindspot Solutions, part of the Adastra Group, has dedicated two decades to the field of artificial intelligence. His journey began in academia, where he first engaged with AI, which subsequently led him to establish his own enterprise. Currently, he offers AI-driven solutions tailored for both emerging startups and established global corporations. With his extensive experience, how does he perceive the evolution of artificial intelligence over the past twenty years?


  • Which milestones have contributed most to the current development of AI?
  • Why is AI now readily available and widespread?
  • What trends should companies watch and where should they be more cautious and wait for further development?

Watch the interview

 

Read the interview

(text has been rewritten and shortened using ChatGPT)

Ivana Karhanová: Now everyone is talking about ChatGPT, which is based on natural language processing models, and AI is enjoying a huge hype. But you told Forbes that you’ve been working with AI for 20 years. What were you working on 20 years ago?

Ondřej Vaněk: When I joined the research group at the university here in Prague, around 2000, it was an active area of research with a multitude of streams of activities and technologies. I was focusing on multi-agent systems, game theory, and a little bit of machine learning. But other people were doing logical reasoning, formal modeling, large-scale simulations, and of course, statistical machine learning as well.

Ivana Karhanová: So what was the first project you worked on?

Ondřej Vaněk: Actually, I was modeling maritime piracy in the Indian Ocean. I was using game theory to understand the behavior between the pirates, the ships, and the patrols.

Ivana Karhanová: How is AI different today, if it’s different at all?

Ondřej Vaněk: It is different in the sense that it has converged almost exclusively with the field of machine learning, and especially large language models, which we have seen recently. Now, what researchers need is huge computational power to be able to train these large language models and then, of course, provide them to the users.

Ivana Karhanová: From your perspective, what were the main milestones in developing AI?

Ondřej Vaněk: So I think from the 70s, when some researchers at MIT said that they needed a year at most to develop a generic reasoning model or technique on par with human reasoning. From that point, it took several summers and winters where there was typically big hype that this technology would solve everything. But then, the researchers realized that it’s not that easy. The same is still happening today. When the first deep learning models appeared, there was a hype that this technology could do pretty much anything: drive cars, understand natural language, and so on. At this point, I believe that the deep learning models, specifically the large language models, are capable of solving a wide variety of tasks. But still, their utilization is limited in some domains, and the deep level of understanding that humans have about the world and processes is not there yet.

Ivana Karhanová: Today, there’s a lot of discussion about using AI in business, especially in larger companies. When you look back at its development over 20 years, how would you describe how much the portfolio of technologies that companies can use has changed over time?

Ondřej Vaněk: Throughout the years, we saw that companies were digitizing and automating a number of things. Part of that was the deployment of some artificial intelligence techniques, typically in some sort of narrow space, a single process, and with, I would say, limited impact. At this point, companies are thinking big. They believe that GPT and related models are actually able to address a wide variety of processes or very advanced optimization technologies, and are able to optimize a wide range of operations. Compared to, let’s say, 15 years ago, we can see that companies are thinking strategically about the technology and how to introduce it in any process or department they have.

Ivana Karhanová: What was the main reason that made AI so accessible and widespread now, even in practical business?

Ondřej Vaněk: I would say that this year, 2023, with the appearance of GPT-3 and GPT-4, marked a milestone when pretty much every human being with access to the internet got access to AI. Everyone can experiment now what AI can do, and that has generated a lot of ideas. This type of AI is now at the level of maturity where it can automate a wide variety of tasks. This is one piece. The second piece is the commercial aspect of that. The AI sits in the cloud, and everyone is connected to the cloud through the internet. The providers of the technology, like Microsoft, Amazon, Facebook, and others, are able to serve all users globally with just a couple of dollars a month.

Ivana Karhanová: So, how has the approach of companies changed with the usage of AI?

Ondřej Vaněk: I would say that they think not in terms of a single point where technology such as AI can help, but they think that their business can be disrupted at this point. They are considering how to keep up with the technology, how to keep up with innovative companies, typically based in Silicon Valley, and how the users and their behavior will change with the introduction of AI.

Ivana Karhanová: You have worked with a wide range of technologies and predictions that said where the world would be in 20 years. Which technologies turned out to be futile and which ones are still with us?

Ondřej Vaněk: So I think it’s notorious for predictions to fail. If we take predictions from 20, 50, 100 years ago, most of them were wrong with respect to timing and being too optimistic. They also neglected some trends that were difficult to predict. Looking 20 years back, we see that things that were with us for a long time, like books, a quite conservative technology, are still with us. This rule of having some technology around for a long time implies that we will have this thing or technology for a long time in the future. On the other hand, new trends are very difficult to estimate;when they will come, where they will appear, and how disruptive they will be. Making predictions about new timing and scientific revolutions is very difficult.

Ivana Karhanová: Can you possibly mention a prediction which was completely wrong?

Ondřej Vaněk: I would say the flying cars are still not there, right? That’s one. But also, even autonomous driving is still limited, and we don’t have it, even though I would say ten years ago, quite a few people were optimistic about that. And education is pretty much the same as 100 years ago. So there are no disruptions as of yet. On the other hand, I think cell phones were predicted to some degree. Like personal assistants having something I can ask, I can connect with the world. That was predicted but with a different timeline.

Ivana Karhanová: Considering your experience, would you dare to say what we can expect from the development of AI in the future? What trends should companies jump on? And what is worth waiting for?

Ondřej Vaněk: So I would say that with the current hype around GPT and large language models, it’s not only hype, but the technology is very useful. In the near years, we will see big competition in this space. The business giants will compete about the market share and the quality and affordability of the models. We will see businesses adopting this technology and finding the limitations of those models. After some period of time—let’s say, two, three years, maybe five —we will need to come up with improved technology, that isnot only evolutionary but some sort of revolutionary shift in which direction that will be. From the technological perspective, it’s difficult to say. We will also see the legal framework catch up with technology, understand what was fed into the models, how they were trained, whether it was legal or not, and how we use them.

Ivana Karhanová: For example, the EU AI Act?

Ondřej Vaněk: Exactly. The EU AI Act is being updated to account for this type of technology. But in general, the EU AI Act is focused on various risks with respect to applying AI. So that would, I would say, stand the test of time.

Ivana Karhanová: Ondra, thank you for sharing your insights with us today.

Ondřej Vaněk: Thank you for inviting me, Ivana.