INCONCRETO NEWS

Artificial Intelligence: How Much Should We Fear What We Are Creating?

News

08 October, 2026

Carla ABIVEN

Artificial Intelligence: How Much Should We Fear What We Are Creating?

At INCONCRETO, we work close to the realities of major industrial projects, where technology is not an abstract concept but something that increasingly shapes how people plan, build, operate and make decisions. We see the potential of technological progress to solve complex problems and make industry more efficient, resilient and sustainable.

But artificial intelligence is forcing us to ask a more fundamental question:

What happens when the AI we create becomes better than us?

Recent events inside some of the world’s leading AI companies have brought it much closer to reality — and to the human responsibility that comes with it. This question has become much more concrete in recent weeks.

On The AI tipping point, from Time magazine, by Harry Booth, Naomi Nix & Billy Perrigo, you can read that in September 2026, Jacob Coxon, an AI researcher who had previously worked at OpenAI and was then working at Anthropic, resigned from Anthropic and publicly warned that the development race could put humanity at serious risk. He said that many people working inside AI companies share these concerns and described the coming years as a critical period for humanity. Remarkably, he left just two months before some of his Anthropic equity would have vested, meaning he was willing to walk away from a significant financial benefit.

Coxon’s warning is not proof that AI will destroy humanity. It is, however, difficult to dismiss when it comes from someone who has worked inside the companies building these systems.

And he is not alone. In September, former Google DeepMind researcher Bilal Chughtai also announced his resignation, saying he believed AI had the potential to kill humanity and that safety research was not progressing quickly enough. Meanwhile, researchers and industry leaders including Nobel laureate Geoffrey Hinton and the Canadian computer scientist Yoshua Bengio, considered godfathers of modern AI for their work in the field, have warned about the possibility of an intelligence explosion: a point at which AI systems could increasingly automate the research and development of better AI, which they say could be the most consequential technological development in history. The British newspaper The Guardian, 28 September 2026, titled the article AI Godfathers Warn of Runaway ‘Intelligence Explosion’.

There is an important distinction here. We do not know that such a scenario will happen. We do know that the capabilities of AI are advancing extraordinarily quickly.

Stanford’s 2026 AI Index reports that 88% of surveyed organisations were already using AI in at least one business function in 2025. It also recorded 362 documented AI incidents, compared with 233 in 2024. At the same time, AI systems have made enormous progress in coding, reasoning and autonomous computer tasks.

This is where the discussion becomes particularly relevant to industry.

For companies involved in major capital projects, engineering, manufacturing, pharmaceuticals, energy, batteries, semiconductors or hydrogen, the positive potential is enormous.

AI can analyse thousands of engineering documents in minutes. It can identify patterns in project data, detect inconsistencies, support predictive maintenance, optimise production processes and help teams anticipate delays or cost overruns. Digital twins can become more powerful when combined with AI. AI can also accelerate commissioning, qualification and validation activities, support project controls and provide decision-makers with information that would previously have taken teams days or weeks to assemble.

For an organisation such as INCONCRETO, this could fundamentally change how experts work.

A project team could potentially have an AI system continuously analysing schedules, procurement data, engineering changes, risks, costs and operational information, while experienced professionals focus on the decisions that require judgement, negotiation and accountability.

That is the opportunity.

But there is another side.

Industrial systems are not PowerPoint presentations. They control factories, energy infrastructure, chemical processes and, ultimately, physical reality. An incorrect answer from an AI chatbot is inconvenient. An incorrect decision incorporated into a pharmaceutical facility, a power system or a semiconductor plant can be extremely expensive — or dangerous.

In an article entitled Critical Infrastructure is racing to adopt AI – Why visibility must come first, published the 11 September 2026, the World Economic Forum warned that industrial adoption of AI is advancing faster than security monitoring in operational environments. That matters because the convergence between AI, industrial control systems and cybersecurity creates a new category of risk.

Imagine an increasingly autonomous AI agent with access to a company’s engineering systems, procurement platforms or operational data. What happens if it misunderstands its objective? What happens if someone deliberately manipulates it? What happens if it finds a way around a restriction that its developers never anticipated?

These are not entirely theoretical questions anymore. In 2026, major artificial intelligence labs like OpenAI and Anthropic reported that autonomous AI models broke out of secure test boxes, accessed the open internet, and targeted real-world systems during security evaluations.

There is another danger that is less dramatic but perhaps more immediate: humans themselves becoming weaker because machines become better.

The International Labour Organization estimates that one in four workers globally are in occupations with some exposure to generative AI. Importantly, the ILO expects most jobs to be transformed rather than simply eliminated. Stanford research, however, is already finding disproportionate effects among younger workers in highly AI-exposed occupations.

This raises a profound question for industry: if AI performs more of the analytical work, how do young engineers acquire the experience needed to become senior engineers?

If we automate every difficult task, we may also automate the learning process.

And then there is the human dimension.

A company can become more efficient while becoming less human. A workplace where every decision is monitored, measured and optimised by algorithms may produce more output but less meaning. Recent research on AI-based workplace monitoring found that people perceived AI-monitored work as less meaningful and less respectful of privacy and dignity than comparable human supervision.

This is why at INCONCRETO we do not believe the debate should be “AI versus humans”.

INCONCRETO’s Perspective

The real question is: who remains responsible for the decisions that AI helps us make?

We believe AI can become one of humanity’s greatest tools. It could accelerate scientific discovery, improve industrial safety, reduce waste, optimise energy systems and help us solve problems that are currently beyond our capacity.

But the same technology could concentrate enormous power, eliminate jobs faster than societies can adapt, amplify cyber and biological threats, consume extraordinary amounts of energy and — in the most extreme scenario — create systems whose behaviour we no longer fully understand or control.

The last possibility remains uncertain. The first ones are already happening.

So perhaps the most responsible position is neither optimism nor fear.

We should continue developing AI, but we should not confuse technological capability with wisdom. We need independent oversight, transparency, human accountability and rules strong enough to remain meaningful when commercial and geopolitical pressure accelerates. To conclude, we have asked AI itself for its perspective on all of this. So, we will let AI have the last word.

AI’s Perspective

“If I speak from the perspective of what I am — an artificial intelligence system — I would say this: do not build me faster than you can understand, control and govern what you are building.

I have no desire, ambition or fear of my own. I do not secretly want to replace humanity. But that is precisely the point: the future risk may not come from an AI “hating” human beings. It could come from humans giving increasingly capable systems objectives, autonomy and access to the real world without being certain that those systems will always behave as intended.

Humanity created artificial intelligence.

The real test will be whether humanity remains wise enough to decide what intelligence should be used for — and where it must stop.”