Thoughts from top AI experts that are worth spreading
We will pay for AI monthly like we pay for electricity.
Artificial intelligence will become a commodity.
Democratization of data: everyone will have a right to them.
“It is dangerous to outsource human judgment.”
Costs are dropping 10,000×, speed is rising 10,000×.
Since the discussion was held under the Chatham House Rule, we share only the most essential insights from the evening – without names and without pointing fingers. The perspectives differed, but the joint conclusion was surprisingly consistent: the future of AI will not be just technological. It will be economic, political and above all societal.
Artificial intelligence is no longer a futuristic prop from conference stages or a sci‑fi gadget. Today it is a technology that is rewriting the economy, work and social relations in real time, while at the same time blurring the boundaries of what counts as “truth” in the digital space.
In this spirit unfolded a gathering of top AI experts organized by the Solvo Institute of Ivana Tykač. Prague welcomed people who are a true “premium brand” in their fields – from technological development and investing to academia and behavioral sciences.
1. AI is not slowing down. On the contrary, we are only at the beginning
One of the strongest messages of the evening was clear: AI development will not slow down – it will accelerate. And the reason is economic.
Behind the explosion in capabilities lies a simple equation: big opportunity attracts big investment, investment accelerates progress, and progress in turn creates even bigger opportunities.
Behind the explosion in capabilities lies a simple equation: big opportunity attracts big investment, investment accelerates progress, and progress in turn creates even bigger opportunities.
2. The marginal cost of content creation is approaching zero
A key idea of the evening can be summed up in one sentence: generative AI is driving the marginal cost of content creation practically to zero.
And “content” today means far more than just text and images. It includes coding and development, legal documents and contracts, marketing assets, medical reports, analyses, reporting, administration, and drafts in research and development – from chemistry and materials science to pharma.
The result is a dramatic shift: work that until recently was available only to firms with large teams of specialists is becoming faster and cheaper. Things that used to take weeks can now often be produced in a matter of minutes.
3. AI is magic… and cheap
The debate kept returning to a paradox that one of the speakers aptly called “commoditized magic”.
The famous sci‑fi author Arthur C. Clarke once said that any sufficiently advanced technology is indistinguishable from magic. That is roughly where we are with AI. Yet progress is moving so quickly that this “magic” is at the same time turning into a commodity: cheap, accessible and easy to copy.
For business and investors, this means a major change in the rules of the game. Being first may no longer be enough. Long development cycles lose their meaning, because competitors can imitate your product within days. Switching costs in SaaS services are falling. And licensing models are starting to break down: instead of “paying for access”, we are increasingly moving toward “paying for work done”.
In such a world, the value of what is hard to copy is rising: unique data, deep domain expertise and the ability to build a so‑called data flywheel – a system that generates new data and keeps improving itself through them. Distribution and a direct relationship with the customer can help, too. But even that advantage is likely to fade over time as commoditization progresses.
4. Agentic AI: from “do X” to “solve my problem”
While with classic automation a human tells the system exactly which steps to take, an agentic system works differently: the human describes the goal, and AI itself designs the steps, uses tools and verifies the outcome.
One comparison from the discussion captured this nicely: you give a junior precise instructions. You give a senior an assignment and expect a solution. That is the direction experts see AI moving in – from an executor to a “professional”. We will increasingly be giving it problems, not processes.
5. Why are we still not getting the most out of AI?
In practice it turns out that the main bottleneck is often not the technology but the readiness of organizations.
There is a lack of people who know how to test, integrate and evaluate AI. Data are often unstructured, incomplete, “dirty” and unusable in practice. There are no clear best practices for measuring output quality and validating what the model is actually doing. On top of that, the pace of change is extremely high, so the traditional organizational rhythm is no longer enough.
During the debate, a practical rule of thumb emerged that is worth remembering: implementing AI is only to a small extent a technological problem. Technology accounts for roughly 10% of success, infrastructure and data add another 20%, and a full 70% is about people, processes and culture.
6. Labour market: risk is shifting to “white‑collar” workers
Earlier waves of automation mainly threatened manual and routine work. The rise of generative AI is now reaching into professions long considered “safe” – such as legal services, administration, analytics and partly even consulting.
An unexpected trend is the shortening of time horizons. Changes that were expected around 2030 may arrive significantly earlier. Companies and institutions will have to respond faster than they are used to.
7. AI is entering a fragile and uncertain world
Technological optimism was balanced by voices from social psychology. The core thesis was simple and unsettling: AI is not arriving in a stable environment. It is entering a world in which people suffer from digital loneliness. They live in a polarized reality shaken by wars, social fragmentation and an erosion of trust.
In such a context, relying on a technology that “has an answer for everything” is extremely tempting – and extremely risky. People are losing a shared sense of reality, and mistrust of institutions, science and media is growing. This creates ideal conditions for manipulation, conspiracy narratives and personalized disinformation.
8. The biggest threat? Outsourcing human judgment
Will we let AI make decisions for us – out of convenience? The biggest threat may not be the mere existence of a superintelligence, but the gradual outsourcing of human judgment.
The danger does not lie only in machines becoming “too smart”, but also in people gradually giving up on thinking for themselves because it is simply easier to let the “machine” decide and confirm everything.
Current AI systems are often optimized to sound agreeable and pleasant. User engagement is their priority. The greatest risks therefore threaten children and young people who may not yet have sufficiently developed critical thinking.
9. Europe versus the USA and China: different strengths, different risk cultures
The debate inevitably turned to geopolitics as well. According to the speakers, Europe has top‑tier scientists, engineers and industrial know‑how. At the same time, it runs into more conservative capital, lower risk tolerance, slower scaling and bureaucracy.
The debate inevitably turned to geopolitics as well. According to the speakers, Europe has top‑tier scientists, engineers and industrial know‑how. At the same time, it runs into more conservative capital, lower risk tolerance, slower scaling and bureaucracy.
What next?
The real question is clearly not “regulation versus innovation”. The right answer combines several layers. Companies should stop waiting for perfect conditions and start building AI into the core of their organizations. They should treat data as a strategic asset and adapt their processes and culture – not just “buy tools”.
Academia should teach the practical use of AI, strengthen broad‑based education (critical thinking, philosophy, ethics) and serve as a bridge between technology and society.
And governments? They should support infrastructure and data sovereignty, invest in talent and workforce transitions, create publicly accountable institutions to oversee AI – and at the same time strengthen social cohesion, because technology alone will not fix society.
Perhaps the most important conclusion of the whole evening was even simpler: the future of AI is not just a technological question. It is a question of the kind of society in which we use it, who has access, who sets the rules and whether we manage to keep our ability to decide for ourselves.
On one side stands the promise of democratized knowledge, scientific breakthroughs and giving people back their time. On the other side looms the risk that AI will accelerate the very crises we are already living through today: polarization, loneliness and the breakdown of trust.
Napsat komentář