For Yoshua Bengio, caution should be the priority. “When the risks are very high, it is crucial to understand before acting, but that is not what happens. The normal wisdom of risk management is to act with caution.” However, opinions differ. He quotes (his friend) Yann Le Cun, who believes that “companies will do what is necessary, otherwise they will go bankrupt.” But for the Quebec researcher, history shows that companies often prioritize their profits to the detriment of protecting the public: “Companies will do what maximizes their profits and put the protection of their public as second priority.” It is therefore essential to supervise the development of AI responsibly to avoid serious consequences
We have many ways to assess risks but few ways to react to them.
What do we do if a test tells us ‘The machine is capable of making very dangerous viruses’, apart from saying ‘stop the system’, we don’t know what to do ,” he explains. Commercial skype database pressure and competition between countries risk ensuring that the race continues despite face of these risks the risks. Research and innovation are advancing at a frantic pace, without taking the time to fully analyze the consequences.
To prevent risks, it is crucial to establish national regulations and international treaties, with clarification of responsibilities:
If there is damage, let us clarify the legislation at the level of civil and criminal liability, so that these companies can be sued if their system causes damage ,” he explains. The fact that companies can be face of these risks held legally responsible could encourage them to adopt more responsible behavior. “ We could even consider forcing them to have insurance. If you manufacture something very dangerous, you need insurance, and the insurer is not biased, he does not want to lose money,” he assures.
Governments also need to step in
We need to go further in the direction of government intervention. Without government intervention, I don’t see how we can get out of this. As AIs become more and more competent, they estonia phone number library are going to be more dangerous. We can’t solve the problem if the government face of these risks doesn’t get involved.” A summit of heads of state is a unique opportunity to discuss solutions to these problems at the highest level. Although researchers have developed techniques to mitigate some risks, no comprehensive solution has yet been found. Policymakers face an “evidence dilemma ,” where risks are often assessed without sufficient data to ensure that mitigation measures are effective.
In 10, 20 years, what kind of world will exist? It is almost certain that we will have machines that are more intelligent than us by then.”
The future of general-purpose AI remains uncertain and depends on the choices of societies and governments. “We are not passive actors , ” insists Yoshua Bengio, stressing that it is still possible to influence its development by establishing regulations and deepening the understanding of the risks. He warns against “the lure of gain that will benefit a few, while ordinary citizens are more likely to lose their jobs .” And he bfb directory calls for remaining optimistic in the face of fatalism. The AI Summit, for example, encouraged the creation of citizen initiatives independent of governments, thus multiplying the spaces for dialogue. According to the researcher, the essential thing is to advance the public debate so that everyone understands the issues and informed collective decisions can be made. “In 10, 20 years, what kind of world will exist? It is almost certain that we will have machines that are smarter than us by then ,” he predicts.