Key insights:
At the World Economic Forum 2026, two of the most important voices in artificial intelligence sat down to discuss what happens after AGI arrives. Demis Hassabis of Google DeepMind and Dario Amodei of Anthropic shared updated predictions, and both believe the timeline is shorter than most people think.
The conversation, moderated by Zanny Minton Beddoes of The Economist, covered everything from job displacement to geopolitical risk. The takeaway is clear. The world is changing fast, and you need to prepare now.
Dario Amodei stuck by his earlier prediction that AI could match a Nobel laureate across many fields by 2026 or 2027. He pointed to coding as the leading edge.
Inside Anthropic, engineers are already saying they no longer write code themselves. They let the model do it, then edit and refine the output. Dario thinks we are six to twelve months away from AI doing most of what software engineers do end to end.
The key driver is the feedback loop. AI builds better AI, which then builds even better AI. The question is how fast that loop can close.
If you are a developer, this should make you think carefully about where your skills sit on the value chain.
Demis Hassabis stood by his prediction of a 50% chance of human-level cognitive systems by the end of the decade. He sees coding and mathematics as easier to automate because the output is verifiable.
Natural sciences are harder. You cannot easily verify a chemical compound or a physics prediction without experimental testing. That takes time.
Demis also believes some missing capabilities remain. Today's models can solve existing problems, but generating original hypotheses and theories is a different level of creativity.
He thinks one or two ingredients are still missing before we get there.
A year ago, Google DeepMind was seen as lagging behind OpenAI. That has changed. Gemini 3 has pushed Google back to the top of the leaderboards, and the Gemini app is gaining market share.
Demis credits the deep research bench and a renewed startup mentality across the organisation. Focus and intensity matter when the pace is this fast.
For you, this means the race is far from settled. The companies leading today may not be leading tomorrow. Skills that work across multiple model providers are more valuable than ever.
Dario Amodei wrote Machines of Loving Grace about the upside of AI. Now he is writing a follow-up about the risks. His framing comes from the film Contact, where humanity must get through its technological adolescence without destroying itself.
Both leaders agree the risks are real. But they are not doomers. They believe these problems can be solved if we work together and move with care.
Dario stood by his earlier claim that half of entry-level white collar jobs could disappear within one to five years. He sees early signs already in software engineering.
Inside Anthropic, he can see a future where they need fewer junior and intermediate staff. The company is thinking about how to handle that responsibly.
Demis agrees the impact is starting. Junior roles and internships are likely to feel it first. Hiring at that level may slow.
But here is the opportunity. Demis said if he were talking to undergraduates today, he would tell them to become unbelievably proficient with these tools. That advice applies to you too.
The people who learn to build with AI will replace the people who fear it. This is exactly why I built the Complete RPA Bootcamp. You go from beginner to pro across Robotic Process Automation, Agentic Automation, Coded Automation, and Computer-Use Agents. Instead of being replaced by AI and automation, you become the one building it.
The labour market has adapted before. Farming went from 80% of jobs to a tiny fraction. Factory work gave way to knowledge work. Each shift created new roles.
Dario's worry is the speed of this transition. Past transitions happened over decades. This one could happen in years.
If AI gets better than humans at most tasks within one to two years, the lag between capability and adoption shrinks. The exponential keeps compounding. Adaptation has limits.
You can either ride the curve or get flattened by it. Learning to build automation systems puts you on the right side.
Dario is concerned about individual misuse, including bioterrorism. He is also worried about nation states, particularly authoritarian governments, gaining access to the most powerful systems.
These are not theoretical risks. As models get more capable, the cost of misuse drops. A small group with the right tools can cause damage that previously required state resources.
Mechanistic interpretability is one answer. Anthropic pioneered this research, which looks inside the model to understand why it behaves the way it does. Both Dario and Demis have neuroscience backgrounds and approach AI safety from that angle.
If we build these systems poorly and race without guardrails, things can go wrong. If we build them carefully, the risks are manageable.
The geopolitical environment has gotten more complex over the past year. The US-China relationship, chip export policies, and tensions between Western allies all shape what AI companies can do.
Both Demis and Dario want international cooperation. Both know it is hard to achieve. The real-world policy debate is messier than the ideal.
Dario was direct on this. He thinks selling advanced chips to China is one of the worst things the US can do. He compared it to selling weapons to a strategic adversary for short-term profit.
His logic is straightforward. Without chip exports, the race becomes between US companies. That is a manageable competition. With chip exports, the race becomes geopolitical, and slowing down for safety becomes impossible.
The current administration takes the opposite view. They argue selling chips binds China into US supply chains. Dario disagrees and thinks the technology is too significant for that calculation.
You may not control chip policy, but you should understand how it shapes the pace of AI development.
Demis sees a real risk of public backlash. Job displacement and fear can drive bad policy, the same way globalisation in the 1990s led to the political situation we have today.
His answer is for the industry to demonstrate clear benefits. AlphaFold is the example he gives. It solved protein folding and accelerated drug discovery. It is hard to argue against that kind of impact.
The industry needs more AlphaFolds. Not just talk, but proof. If people see real cures and real progress, the political case for letting AI develop becomes stronger.
For you, this means the companies and projects you back should be doing more than chasing chatbots. The real value is in solving hard problems.
Demis raised something most economists miss. Jobs are not just about money. They give us meaning, structure, and identity.
If AI handles most economic work, we may end up in a post-scarcity world. Distributing that wealth is one problem. Finding purpose is another.
His answer is that humans already do many things without economic motivation. Sports, art, exploration, curiosity. These will become more sophisticated, not less important.
He also mentioned exploring the stars. That is not a small idea. If AGI removes the bottleneck on science and engineering, the solar system becomes our playground.
But this is five to ten years away, not fifty. You need to think about your role now.
If Demis and Dario are even partially right, the next few years will reshape every industry. Software, science, government, and education will all look different.
You have two choices. Wait and see what happens, or learn the tools that will define the next era.
Coding is the first field being automated. That sounds scary, but it actually opens a door. The people who understand how to direct AI systems will be more valuable than the people who write code line by line.
Robotic Process Automation is the foundation. It teaches you how to break down business processes, design workflows, and integrate systems. These are skills AI does not replace, it amplifies.
Agentic automation takes this further. Instead of rule-based bots, you build agents that reason, plan, and act. This is exactly the direction Anthropic and Google DeepMind are pushing.
The Complete RPA Bootcamp covers all of this. RPA, agentic automation, coded automation, and computer-use agents. It is built for people who want a future-proof career.
Based on the conversation, here is what will matter:
None of these are about typing code faster. They are about being the person who turns AI capability into business value.
You do not need a computer science degree. You need a structured path and the willingness to put in the work.
Start with these steps:
The bootcamp gives you all of this in a structured format. You move from absolute beginner to a working Automation Developer who can compete in the new economy.
Demis Hassabis and Dario Amodei agree on more than they disagree. AGI is coming. The exact year matters less than the direction. The risks are real, but so is the upside.
The Fermi paradox came up at the end of the conversation. The idea that we should see signs of alien civilisations and we do not. Demis thinks the great filter is behind us, not ahead. He believes we get to write what happens next.
That is the right frame. The future is not something that happens to you. It is something you build.
If you want to see the full debate and watch two of the smartest people in AI lay out their views in their own words, watch the embedded video below from the DRM News YouTube channel. It is well worth your time, and it will shape how you think about the next five years.