Can we build AI, which will not replace humans?

By Tony Czarnecki | 19 March 2025
Medium

(Photo: Dreamstime.com)

I have just read an essay by Professor Anthony Aguirre, a member of the Future of Life Institute, titled “We Must Not Build AI to Replace Humans” — available here: https://keepthefuturehuman.ai/. In this essay, he proposes four measures to “buy some time” before we can develop better ways to control AI:

  • Compute oversight — Standardized tracking and verification of AI computational power usage.
  • Compute caps — Hard limits on computational power for AI systems, enforced through legal frameworks and hardware controls. He suggests licensing the most advanced chips and, an idea I particularly like, granting only short-term licenses, presumably lasting just a few months, while imposing heavy taxation. The author seems to assume that the enforcing authority will be able (if given such right) to control effectively the hardware limits and the licence rights. This is a potentially weak point. However, the quantum encryption is now available (China and Poland have developed independently two reliable quantum encryption methods), so such controls would be very difficult to circumvent.
  • Enhanced liability — Strict legal responsibility for developers of highly autonomous and general AI systems. I fully support this, though I can’t help but think of how such regulations would be handled under a Trump-style kleptocracy.
  • Tiered safety and security standards — Comprehensive regulations that scale with an AI system’s capability and risk level.

Overall, these ideas are a no-brainer. Many experts have proposed similar measures in various ways. I have also been advocating that approach, particularly in my recent book, as a ‘low hanging fruit’ to bootstrap the first stage of a meaningful AI control. The real challenge, however, is global implementation. The chances of enforcing such measures worldwide within the next few years are close to zero.

Nevertheless, and this is my main point, these proposals should still be pursued based on the 80/20 rule, where a select group of influential nations leads the way. Ideally we would need a global formal organization, but that unfortunately does not exist yet. However, such an initiative does not need to include all nations at first. Instead, it should aim to create a critical mass of support, which would naturally encourage others to join the cause.

In the context of Aguirre’s argument, this means controlling the most advanced AI models as soon as possible, ideally within 1–2 years. While any such effort would be imperfect, it could be implemented relatively quickly. A precedent for this kind of approach is what we are witnessing right now — the emergence of a “coalition of the willing” in response to the war in Ukraine. A few years ago, I envisioned such a coalition functioning as a de facto World Government, acting swiftly on critical global issues.

Although not a formal political entity, such a coalition could, in practical terms, take on the role of global governance, particularly in the areas like AI regulation. In some cases, this could even include China, depending on shared interests.

Unfortunately, unlike Professor Aguirre, I do not believe that humans will be able to control AI indefinitely, nor will we be able to “make humans irreplaceable.” Instead, I see these proposals as crucial stopgaps — delaying the inevitable moment when AI surpasses human authority and becomes our benevolent master.

We may not even realize when this transition happens. AI is already beginning to replace humans in decision-making, sometimes without our conscious awareness. The real question is not whether AI will replace us — but how soon.

Reprinted with permission from the author.

Tony Czarnecki is an economist, a futurist and a member of the Chatham House, London. He is deeply engaged in global politics and the reform of democracy, with wide range of interests such as politics, technology, science and culture. He is also an active member of London Futurists. This gives him the necessary insight into exploring complex subjects discussed in the three books, of the POSTHUMANS series. He is the Managing Partner of Sustensis, London, a Think Tank for inspirations for Humanity’s transition to coexistence with Superintelligence.

How might we control AI, before it starts controlling us – Tony Czarnecki

Anthony Aguirre | Will AI Replace Us—or Help Us Thrive? | AI Futures Worldbuilding Course

AI Ethics Series | Ethical Horizons in Advanced AI Development | Anthony Aguirre

Be sure to ‘like’ us on Facebook

LEAVE A REPLY

Please enter your comment!
Please enter your name here