From Elon Musk to Geoffrey Hinton, many experts have accepted the possibility that artificial intelligence will destroy humanity. It is true that some in a greater proportion than others, but that risk seems to be there. However, how much truth is there in that idea and how much is it influenced by science fiction? A study has set out to resolve the issue.
To do this, it has decided to study in depth the moment that artificial intelligence is currently going through and try to answer the million-dollar question: Does this technology really pose a threat to the world as we know it? His answer is clear and unadorned: no.. However, they have added some fine print: there are other risks.
No, AI will not wipe out humanity
We have all imagined a similar futuristic scenario at some point, or have seen it in a movie. An artificial intelligence, often in the form of a robot, decides that human beings are a problem, a nuisance or whatever, and does everything in its power to eradicate them. A little bit like Terminator o Matrix. Can something like this happen?
Since the rise of ChatGPT and other similar AI chatbots, many experts have argued that yes, the threat is not negligible. Now, however, a study says just the opposite: it will not be like that. Only this second one is not based on intuitions or hunches, but on a critical and thorough analysis, as those responsible have shared on arXiv.
To find out if there could really be any real possibility of a practically apocalyptic scenario, they have reviewed the arguments that support catastrophic predictions; They have then compared them with the empirical evidence available between 2023 and 2025. The result? That There is nothing to suggest that such a tragedy could occur because of AI.
Or what is the same, no continuous recursive improvement, no autonomous strategic awareness or uncontrollable misalignment. There is nothing of that in the current models that, according to those responsible for the investigation, They are powerful statistical artifacts, but nothing more. According to them, there are no arguments of any kind that foresee extreme risks.
The trees, the forest and the AI
The conclusion of the study, therefore, leaves no room for doubt: those who talk about a threat from AI to humanity, in the style of a science fiction movie, do so more out of ideology or some particular interest than because of scientific evidence as such. In fact, they suggest that perhaps the intention of many is to use their imagination to hide threats that are not real.
For example, excessive automation or economic problems. Both would have obvious consequences, such as massive job loss. (which is already happening) or greater inequality between the very rich and the poor. That, they say, is something that could happen. Or rather, it is already a reality.