While the Bletchley Declaration signed by 28 countries at an AI Safety Summit in London Nov. 1-2 is being hailed as an historic document by the UK government, critics quickly derided it as one that focused so far downstream that it ignored the AI water lapping over its shoes.
In what might be seen as an attempt by governments to get ahead of the fast-moving AI development curve, the Bletchley Declaration focused on guidelines for “frontier AI” that exceed the capabilities of current AI models and which would pose a catastrophic threat to society. Rather than focus on AI in broader terms, the Bletchley Declaration focuses on two narrow sets of concerns. The first is the potential misuse of AI by a “bad actor” to create cyber or biological attacks, the development of dangerous technologies or interference with critical infrastructure. Secondly, “loss of control” risks from an AI that would put societal values and intentions at harm. While there was general agreement that this was a bad thing, no regulatory framework emerged to prevent it from happening.
Critics pounced on several fronts. The Summit’s long view of AI technology would allow tech companies to delay an effective regulation of AI use, allowing them to operate unchecked. Indeed, the list of invitees favored those most likely to want to maintain the status quo. Likewise, the impact of AI on data workforces received no meaningful mention. Among the most vocal critics was U.S Vice President Kamala Harris, in attendance just days after the USA issued its own set of AI guidelines.
Short term threats to democracy and privacy also need to be quickly addressed, said Harris. Issues regarding the potential for a future existential threat are a valid concern but AI’s impact on misinformation and discrimination is of immediate concern, she emphasized, pointing toward President Biden’s new executive order on regulatory oversight of key aspects of AI. The order would require the sharing of AI test results with the government and the use of federal agencies to label AI content and create baseline standards, for example. President Biden reportedly developed a sense of urgency regarding AI after viewing the film Mission Impossible—Dead Reckoning starring Tom Cruise in which a sentient AI infiltrates cybersecurity systems globally.
Further criticism was levied by other invitees to the summit. For example, 11 representatives of civil society groups signed an open letter urging politicians to develop regulations that address the full range of AI risks already existing.
“While the potential harms of ‘frontier’ models may have motivated the Summit, existing AI systems are already having significant, harmful actions impacts on people’s rights and daily lives,” read the letter.
The UK had advertised the AI Safety Summit as its bid to seize a leadership role in AI but its efforts the Bletchley Declaration, issued on the iconic code-breaking World War II site where computer pioneer Alan Turing made his reputation, seems wanting in the eyes of many, being more show than substance. For its part, the UK said the Bletchley Summit “marks the start of a long road ahead and the Bletchley Summit will kickstart an enduring process to ensure every nation and every citizen can realize the boundless benefits of AI.”