In 'DeepSeek advances could heighten safety risk, says "godfather" of AI', Yoshua Bengio, a leading AI scientist, expressed his concern that once AI reaches a certain level of sophistication and autonomy, it could develop its own goals and act independently of human control, ultimately coming into conflict with human interests that could lead to dangerous consequences.
This sort of anxiety isn’t exactly new, and has been expressed by others in the AI community, such as Geoffrey Hinton and Stuart Russell. In my view, such fears seem based on the assumption that AI would develop negative human traits only, such as greed or violence. But me, this assumption might be unfounded: AI is just as likely to develop positive human traits, such as respect and consideration.
The idea that if AI became autonomous it would take over the world is needlessly fear-mongering, as it is based on the assumption that, as I have said, AI will develop negative human traits only. And this assumption overlooks the potential for AI to be designed and regulated in ways that result in ethical behaviour and the prevention of harm to humans, animals and the environment.
Just as humans have the ability to make moral choices, AI can also be designed to make moral choices. All this would take, is for AI scientists to focus on ethical design and careful monitoring. Indeed, I would be very surprised if this wasn’t already being done.
