An open letter co-signed by employees of OpenAI and DeepMind calls for an extension of the whistleblower status that allows members of large companies to disclose secret information when the future of humanity is at stake. “How can we not subscribe to it, in the face of the announced catastrophe?” asks our columnist Jean-Gabriel Ganascia.
This article is taken from the monthly Sciences et Avenir – La Recherche n°931, dated September 2024.
Whistleblowers, whistle-blowers – of whistle, whistle, and blowblow in English – or carillonneurs draw attention to reprehensible facts or those contrary to the general interest, so that they can be investigated and beneficial measures taken as quickly as possible.
Whistleblower status recognized since 2016
Clearly, when they concern attacks on people’s health or the deterioration of the environment, such revelations benefit everyone. They must be encouraged, despite the obstacles that arise. To prevent their authors from suffering from those they implicate, laws protect them in both North America and Europe, particularly in France, since the status of whistleblower was recognized in 2016.
On June 5, 2024, an open letter co-signed by current and former employees of OpenAI and DeepMind, and sponsored by three industry figures, Geoffrey Hinton, Yoshua Bengio, and Stuart Russell, denounced the hidden dangers of artificial intelligence techniques. It took the opportunity to call for an expansion of the whistleblower status that would allow employees of large companies to disclose secret information when the future of humanity is at stake.
Read alsoArtificial intelligence suffers from a lack of trust
70% chance that AI will destroy or irreversibly harm humanity?
How can we not subscribe to it, in the face of the announced catastrophe? Except that through repeated scares, their authors lose all credibility. What are these secrets that are surfacing without our knowledge? They say nothing, except that they will cause a cataclysm.
One of the signatories, former OpenAI employee Daniel Kokotajlo, said in a statement made at the end of May that New York Times70% chance that the general artificial intelligence produced by OpenAI and DeepMind will destroy or irreversibly harm[…]
- sciencesetavenir.fr
