Biden's AI Directive: Exploring the Implications of the Executive Order

0:00In an effort to establish a more secure foundation for AI technologies, US President Joe Biden has released an executive order mandating AI developers to notify the federal government and present results of all safety safety. before a public launch. I think right now we're seeing, of course, like the rapid surge in AI. ChatGPT has obviously propelled this a lot, but it has prompted international discussion about the need for checks and balances in algorithm-driven systems. I think in a previous meeting in May, a bunch of G7 leaders kind of pinpointed their primary areas of concern. And I think today, the G7 nations have come to a consensus on some core guiding principles and introduced a quote-unquote voluntary code of conduct aimed at AI developers.

0:47So while they have done this, I think the global community is also kind of turning its focus on AI governance. The United Nations unveiled a new kind of board last week to go into this matter. The UK is gearing up for a worldwide summit on AI governance. And I think this is happening at Bletchley Park, where they have Kamala Harris is scheduled to address the attendees there. The US has received some criticism in this regard. Kamala Harris was named as kind of the head of AI for the United States. Of course, she has no background in this area. So there was criticism there. But the Biden Harris administration has also prioritized AI safety.

1:28I'm collecting essentially like they got voluntary commitments from a bunch of major AI players, OpenAI, Google, Microsoft, Meta, Amazon. and I think this is kind of this that effort there that they put forth was kind of seen as kind of like the precursor to the current executive order they've just launched so like you have to give them some credit that they are definitely thinking about this they're definitely talking about this they're pushing forward the conversation in this space they do have of course criticism of credibility in this space obviously Kamala Harris isn't leading I would hope the all of the legislation around this I'm sure they have experts in there but you know it just wasn't great optics to kind of name her instead of someone that's more that has a lot more experience in this area.

2:11But that aside, give them credit for, you know, pushing the conversation forward in this direction. The executive order says essentially that, quote, most powerful AI systems. So that's who they're addressing. Developers must disclose their safety test results to the U.S. government. When they say most powerful AI systems, what they're really talking about is Google, Microsoft, Meta, Amazon. Interesting. It's hard to determine exactly like what classifies, right? A new startup comes out with something more powerful than OpenAI, for example ,'s ChatGPT. Would they get lumped in with that? How would we know they're the most powerful AI system?

2:46There's a lot of small startups. I think they're trying not to step on their toes and stifle those startups. If it gets to a point where that happens, then it would appear that, you know, OpenAI would have received some regulatory, a regulatory moat because they can kind of shut down smaller startups that might be able to threaten them. There's all of those concerns. The executive order essentially aims to quote unquote protect Americans from the potential risks of AI systems. And I think by linking the AI safety guidelines to the Defense Production Act of 1950, essentially they're trying to address any foundation model that could jeopardize national security, economic stability, or public health.

3:24Again, really hard to say exactly what that would look like, what those are. So that is definitely something that is going to get litigated and discussed in the future. The order's reach, I think, doesn't really end there. There's plans in place to have a new mechanism to certify AI's safety. So the National Institute of Standards and Technology, NIST, has been assigned the role of creating standards for comprehensive red teaming tests prior to any AI product release. And, you know, it's interesting. We're seeing, of course, OpenAI launched a big, huge red teaming, red team testing department earlier this year.

4:02And I know that OpenAI has been consulted in kind of discussions around all this. So I wonder if they were, you know, essentially knew this was coming down the pipe and were doing that proactively ahead of time for this particular piece of executive order. So despite, I think, its breadth, a lot of critics suggested that the order might actually lack the necessary enforcement power. So while it underscores areas like equity and civil rights and highlights the risks of AI in sectors like healthcare, justice, and housing, a lot of the executive order kind of leans on recommendations and best practices rather than enforcing a specific mandate.

4:36So they say like, hey, you guys should do X, Y, and Z. This is best practice. But they're not exactly giving straight out rules. So it has gotten some criticism for that. I think the big issue of data privacy is acknowledged, especially as AI facilitates, you know, large scale extraction of personal data. The executive order calls on Congress to enact, quote, bipartisan data privacy legislation, essentially to safeguard American citizens information, right? That makes sense. Europe is on the verge of enacting a comprehensive AI regulation. And I think that is really kind of signaling a global move to manage the potential upheaval that AI possibly could bring.

5:14You know, that's what they're saying. And this is, you know, like, in my opinion, the impact of AI really is kind of reminiscent of the Industrial Revolution. There are some major shifts and changes that we're going to go through in the next five years. I think whether President Biden's executive order is going to effectively regulate AI giants like OpenAI, Google, Microsoft, and Meta is still to be told. But I mean, I do think it's promising that they are looking at this. They are, you know, taking steps in this direction. Of course, there's concerns about like overregulating too soon, stifling innovation.

5:47But, you know, I've heard a lot of arguments for regulation as well, where people are saying, you know, when there is regulation, people know what they can and can't do. and putting down those clear lines sometimes helps spur innovation because people will invest in areas where they're like, oh, look, there's regulation saying you can do this. I'll put money in and it spurs innovation. So there's arguments on both sides of that. But I think it's going to be interesting to see how this plays out and what impact this has in the future.