On 23 January 2026, Anthropic CEO Dario Amodei made a prediction at the World Economic Forum in Davos that sent ripples through the tech industry: AI could be doing "most, maybe all" of what software engineers do within six to twelve months.
The statement, made during a panel titled "The Day After AGI", was immediately contested by other AI luminaries on stage, setting up a real clash between some of the most influential figures in artificial intelligence.
For UK businesses that hire developers, commission websites, or build digital products, this raises an urgent question: what should you actually do with this information?
What Amodei Actually Said
The headline is alarming, but context matters. Here is Amodei's full statement, as reported by Fortune:
"I have engineers within Anthropic who say I don't write any code anymore. I just let the model write the code, I edit it, I do the things around it. I think we might be 6 to 12 months away from when the model is doing most, maybe all of what SWEs do end-to-end."
- Dario Amodei, CEO of Anthropic, at Davos 2026
Note what he is describing: engineers who no longer write code themselves but who direct AI to write it for them. This is not replacement. It is transformation. The engineer is still there, still employed, still essential. Their job has changed from writing code to directing and validating AI-generated code.
Amodei went on to acknowledge significant caveats: "I think there's a lot of uncertainty, and it's easy to see how this could take a few years." He listed chips, chip manufacturing, and model training time as constraints preventing AI from "closing the loop" entirely.
Other AI Leaders Strongly Disagree
Amodei was not speaking in a vacuum. On the same Davos stage, two other AI titans offered starkly different views.
Demis Hassabis, the Nobel Prize-winning CEO of Google DeepMind, said current AI systems are "nowhere near" human-level artificial general intelligence. He estimated only a 50% chance of achieving AGI within the decade, and emphasised this would require "one or two more breakthroughs" beyond existing technology.
Yann LeCun, Meta's former Chief AI Scientist who recently departed to launch his own $3.5 billion AI startup, went further. He declared that large language models will "never achieve human-level intelligence" and criticised the industry for being "completely LLM-pilled." LeCun believes an entirely different approach (what he calls "world models" that understand physical reality) is necessary.
This is not a minor disagreement. Three of the most respected figures in AI, with decades of combined experience, hold completely incompatible views on how close we are to AI that can truly replace human cognitive work.
What Most Coverage Gets Wrong
Headlines screaming "developers have 6 months left" miss the point entirely. Even Amodei's own engineers, at the company building Claude (one of the most capable AI coding assistants), are not unemployed. They have changed what they do.
The shift Amodei describes is from writing code to directing AI that writes code. This is a real transformation, but it is not elimination. Consider what this actually means in practice:
| Traditional Developer | AI-Augmented Developer |
|---|---|
| Writes code line by line | Prompts AI, reviews output, iterates |
| Deep expertise in syntax and language quirks | Deep expertise in problem definition and architecture |
| Speed limited by typing and debugging | Speed limited by clarity of requirements |
| Value from technical implementation | Value from judgment, taste, and direction |
Anthropic itself demonstrated this with Claude Code. Their new Cowork product "was written in like a week and a half, almost entirely with Code," according to Amodei. But it still required human engineers to define what to build, evaluate whether it worked, and make decisions AI cannot make. The following day, Anthropic confirmed that Claude Code had hit a billion-dollar revenue run rate, with non-coders driving a significant share of adoption.
What UK Businesses Should Actually Do
If you are a UK business owner who hires developers, commissions web design work, or relies on custom software, here is the practical takeaway:
1. Do Not Delay Projects Waiting for AI
The worst response to Amodei's prediction is paralysis. Even if AI does transform development within 12 months (which Hassabis and LeCun doubt), your business needs to operate now. A website that brings in customers today is worth more than a hypothetical AI-built one next year.
2. Expect Developer Productivity to Increase
AI coding tools are already here. Developers using Claude, GitHub Copilot, or similar tools are more productive than those who do not. When hiring or commissioning work, ask whether the developer uses AI assistance. Those who do can deliver more value in less time. If you are evaluating the options yourself, our sister site's Claude vs ChatGPT comparison for coding and creative writing can help you understand the practical differences.
3. Value Judgment Over Typing Speed
The shift Amodei describes elevates certain skills: understanding your business requirements, translating them into clear specifications, evaluating whether a solution actually works. These are precisely the skills that distinguish a good web design partner from a cheap one.
4. Invest in Your AI Visibility Now
Regardless of who builds your website (human, AI, or hybrid), it needs to be visible to the AI systems that increasingly influence how customers find businesses. Our AI Visibility Checker shows how ChatGPT and other AI assistants currently perceive your business. This matters whether Amodei's timeline is accurate or not.
The Real Risk Is Not What You Think
The real risk for UK businesses is not that developers disappear overnight. It is that competitors who embrace AI-augmented development will move faster, iterate more quickly, and deliver better digital experiences while you wait to see what happens. Meanwhile, the platforms themselves are evolving rapidly. ChatGPT is introducing ads while DeepMind warns about trust. These changes affect how businesses interact with AI.
ServiceNow CEO Bill McDermott, speaking at the same Davos conference, offered a more grounded perspective. His company is not laying off developers. It is retraining them to manage AI agents and work alongside automated systems. "I only wanted nines and tens," he said, referring to keeping his best talent and helping them adapt.
This is the mature response: neither panic nor denial, but practical adaptation.
What to Watch
Amodei's prediction gives us a testable timeframe, and early results are already in. In February, Anthropic launched Claude Cowork plugins for legal, finance, and sales workflows, wiping $285 billion off software stocks in a single day. By late 2026, we will know whether AI can actually handle "most, maybe all" of software engineering end-to-end. Here is what to monitor:
- Claude Code adoption: How many developers actually stop writing code?
- Startup founding patterns: Are non-technical founders successfully building products with AI alone?
- Enterprise deployments: Do large companies reduce developer headcount or redeploy them?
- Quality metrics: Does AI-generated code match human quality for complex, production systems?
We will report on these developments as they unfold. For now, the wisest course is informed action, not waiting.
Frequently Asked Questions
What exactly did Anthropic's CEO say about software developers?
At the World Economic Forum in Davos on 23 January 2026, Dario Amodei said: "I think we might be 6 to 12 months away from when the model is doing most, maybe all of what SWEs [software engineers] do end-to-end." He noted that some Anthropic engineers already direct AI to write code rather than writing it themselves.
Do other AI experts agree with this prediction?
No. Google DeepMind CEO Demis Hassabis said current AI is "nowhere near" human-level intelligence and gave only 50% odds of AGI within a decade. Former Meta AI chief Yann LeCun stated that large language models will "never achieve human-level intelligence" and called the industry "completely LLM-pilled."
Should I wait to commission website work until AI can do it?
No. Delaying projects in anticipation of AI improvements means lost business today. Even if Amodei's timeline is accurate, the transformation he describes still involves human direction and judgment. Good web design is about understanding your business and customers, skills AI does not replace.
Will all software developers lose their jobs?
Amodei's own description suggests transformation, not elimination. His engineers at Anthropic still work there. They direct AI rather than write code manually. The role has changed from typing code to defining requirements, evaluating output, and making architectural decisions that AI cannot make independently.
What should UK businesses do right now?
Continue with digital projects that serve your business. When hiring developers or agencies, ask whether they use AI tools, as those who do are typically more productive. Invest in AI visibility so AI systems recommend your business. Focus on clear requirements and good judgment, which remain essential regardless of who or what writes the code.
Is Your Business Ready for AI-First Discovery?
Whether developers are writing code or directing AI to write it, your digital presence needs to be visible to AI systems. Check how ChatGPT and other AI assistants currently describe and recommend your business.
Check Your AI Visibility
