The proof is in the pudding
Can we trust the AI industry?
Earlier today the White House announced that seven technology companies will adopt voluntary guidelines to govern Artificial Intelligence. The urgency by the Biden administration in setting the tone is a welcomed development since the US, paralyzed by congressional deadlock and conflicting interests, has been falling behind in driving policy in this space. Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI make up the initial list.
The initial set of safeguards the industry is voluntarily agreeing to represent a solid framework with a focus on safety, security, and trust including:
Security testing of A.I. products, including information sharing with governments and others attempting to manage technology.
Research on the risks of bias, discrimination, and privacy violations.
Implementation of watermarks or other means of identifying AI-generated content vs. organic, which is a key element in the fight against disinformation.
On-going reporting of capabilities and limitations, including security risks and evidence of bias.
Deployment of AI tools to tackle society’s biggest challenges (like climate change).
This addresses the concerns and adopts some of the measures that I proposed last April here in this blog.
The agreement, however, does not have the force of law and relies, instead, on a view that corporations have incentives to portray their products as trustworthy fostering consumer adoption. While this is good, history shows that profit and public interest are not aligned and they tend to diverge as competition intensifies. Today’s events sound like a prelude to upcoming legislation with the industry trying to stay ahead of it.
The United States remains the largest and most important AI market, with most leading AI companies based in the US. However Chinese competitors like the BAT (Baidu, Alibaba, Tencent) and startups like SenseTime, Cambricon, and Cloudwalk make China the second AI market in the world, operating with the support of the government, with access to massive amounts of data and not guided by Western ethical standards. Balancing regulations that protect the public interest with supporting thriving product innovation here in America continues to be the main goal.
As we head into the next chapter of this discussion, the proof is in the pudding: is the industry acting in good faith or are these key players engaging in early regulatory capture driving the agenda to be regulated the way they want to be regulated? Time will tell.