A number of prominent AI researchers, including Elon Musk, have signed an Open Letter calling for AI labs around to stop developing large-scale AI systems. They fear the “profound dangers to society and humanity” this software presents.
The nonprofit Future of Life Institute published the letter. It noted that AI labs are in an “out of control race” to create and deploy machine-learning systems “that no one, not even their creators, can understand, predict, nor reliably control.”
“We ask all AI labs immediately to halt for at least six months the training of AI system more powerful than GPT-4.”
The letter states that all AI labs should immediately suspend for at least six months any training of AI systems stronger than GPT-4. This pause should be made public, verifiable and inclusive of all relevant actors. If such a pause is not possible to be implemented quickly, governments should intervene and establish a moratorium.
Yuval Noah Harari is an author. Jaan Tallinn is the Skype co-founder. Andrew Yang is a politician. There are also a few well-known AI researchers, CEOs and leaders like Stuart Russell, Yoshua Benjamin, Gary Marcus and Emad Mostaque. Here is the complete list of signatories. However, new names should not be taken lightly as some reports have suggested that they were added to this list in a joke. OpenAI CEO Sam Altman is partially responsible for the current AI race dynamic.
It is unlikely that the letter will have any impact on current AI research climate. This has seen tech companies such as Google and Microsoft rush for new products and ignore previously-avowed ethics and safety concerns. It is however a sign that there is growing opposition to the “ship it now, fix it later” approach. This opposition could possibly make its way into politics for consideration by actual legislators.
OpenAI, as noted in the letter has suggested that future AI systems may need to be subject to an “independent review” in order for them meet safety standards. Signatories agree that it is now.
They write that “AI labs should take this time to develop and implement a set shared safety protocols for advanced AI development and design. These protocols will be rigorously audited by outside experts and monitored by them.” These protocols should be able to ensure that all systems complying with them are safe beyond reasonable doubt.
Another photo featured several photos of Donald Trump being taken by police officers. These images circulated for days after he claimed that he would not be arrested without any evidence. A Truth Social post by the former US president featured an AI-generated photo of Trump kneeling in prayer.
According to tech leaders, AI labs should immediately stop training AI systems that are more powerful than HTML4 for at most six months while they create safety protocols that would have an independent panel overseeing them.
The letter states that these protocols “should ensure that systems adhering them are safe beyond any reasonable doubt.”
“This doesn’t mean that AI development should be halted, but rather a step back from the race to create ever-larger and more unpredictable black-box models with emerging capabilities.”
The letter called for new regulators and oversight, public funding for AI safety research, liability in the event of AI-caused harm, and other measures.
After an AI-generated photo circulated showing Musk allegedly holding hands with Mary Barra (head of GM’s car company), Musk responded on Twitter.