The AI Daily Brief: Artificial Intelligence News and Analysis - OpenAI Launches Education and Safety Initiative

Episode Date: January 30, 2024

OpenAI has launched an education and safety initiative with Common Sense Media. Also, Biden's Executive Order on AI bears fruit. ABOUT THE AI BREAKDOWN The AI Breakdown helps you understand the most important news and discussions in AI.  Subscribe to The AI Breakdown newsletter: https://theaibreakdown.beehiiv.com/subscribe Subscribe to The AI Breakdown on YouTube: https://www.youtube.com/@TheAIBreakdown Join the community: bit.ly/aibreakdown Learn more: http://breakdown.network/

Today on the AI breakdown, we're looking at a new OpenAI education initiative, the White House reporting on the AI executive order, and so much more. The AI breakdown is a daily podcast and video about the most important news and discussions in AI. Go to Breakdown.network for more information about our YouTube, our Discord, and our newsletter. Hello, AI friends, quick note today. Next couple days are going to be a little bit weird. There is a lot going on, a bunch of meetings and travel.

And so I'm going to be doing extended brief episodes with no main part of the So basically a longer headline kind of story. It'll be a little bit shorter than our normal episodes, but then on Thursday we will be back to normal. But I figure this was better than just abandoning you guys entirely, so I hope you enjoy it. Welcome back to the AI breakdown brief. All the AI headline news you need in around five minutes. We kick off today with a new partnership between OpenAI and Common Sense Media.

So basically, this is part of a broader set of initiatives, I think, to try to make AI ready for all sorts of different types of audiences. The intended target here is young people, teens, who through this partnership, OpenAI and Common Sense Media are trying to help them understand how to use AI in a way that is both safe and effective. At an event yesterday, OpenAI CEO, Sam Altman said, we want to figure out how to make this tool safely and responsibly and broadly available to teens and people who are going to use it as part of their educational experience. Now, common sense makes well sense as a partner for this because their whole focus is making technology safe and accessible to kids, and they had previously been working on an AI rating and review system

designed to help parents, children, and educators actually understand the risks and opportunities of the technology. Now, one thing that's not clear is exactly what the output of this will be. From CNBC, quote, OpenAI and Common Sense didn't say how LLMs will be tweaked to help aid educators or teens. Altman said LLMs, customized for educational purposes, could help teens, quote, who want to learn about science or learn about biology. He continued, I don't think we know exactly how people are going to want to use it. Now, overall, I think this is a totally fine initiative, but frankly, I'm not terribly worried about teens figuring out

how to use a incredibly useful new technology like this. In fact, I am firmly in the camp that whether educators accepted or not, ChetGBT and the like have already completely upended and changed and frankly killed homework forever as we used to know it. However, I'm also in the camp that that's a good thing and is going to lead to better educational outcomes as we move past strategies that were useful 100 years ago. Next up, we're starting to get more information about how some of the provisions of the

Biden executive order on AI is actually going to be implemented or find its way into policy. Yesterday, the White House released a note that said, Today, Deputy Chief of Staff Bruce Reed will convene the White House AI Council, consisting of top officials from a wide range of federal departments and agencies. Agencies reported that they have completed all of the 90-day actions tasked by the EO and advanced other vital directives that the order tasked over a longer time frame. Taken together, these activities marked substantial progress in achieving the EO's mandate to protect Americans from the potential risks of AI systems while catalyzing innovation in AI and beyond.

Now, the most notable of these new policies or updates was the one we reported on yesterday, where the Defense Production Act is now being used, as the White House puts it here, compel developers of the most powerful AI systems to report vital information, especially AI safety test results to the Department of Commerce. Another widely reported update is a, quote, draft rule that proposes to compel U.S. cloud companies that provide computing power for foreign AI training to report that they are doing so. Basically, one of the vectors for monitoring and surveillance that the U.S. government is using is how people are using cloud computing resources and they can use the fact that most of those

cloud computing resources are housed in the U.S. as a way to keep tabs. Other initiatives that have launched over the last 90 days include a pilot of the National AI Research Resource, that's something that we reported on last week. A, quote, AI talent surge to accelerate hiring AI professionals across the federal government, including through a large-scale hiring action for data scientists, begin the Educate AI initiative to help fund educators creating high-quality, inclusive AI educational opportunities at the K-12 through undergraduate levels, and a number of other initiatives as well.

Basically, this is a little bit of a progress report, and the White House is saying we are moving swiftly on all these things. Although, of course, the big question is when we will actually see comprehensive legislation from Congress to match this, although given that it's an election year and things are in such turmoil right now, I'm not sure that that's totally realistic. Indeed, that was the frame that Reuters went with. White House AI Council meets Monday as legislative action stalls. Reuters writes, the White House said nine government agencies, including defense, transportation, treasury, and health and human services, submitted risk assessments to the

Department of Homeland Security required under Biden's order. At the same time, efforts in Congress to pass legislation in Congress, wow, great writing, addressing AI have stalled despite numerous high-level forums and legislative proposals. Now, moving over to markets for a moment, as we've discussed over the last couple weeks, meta is clearly making some big moves in their AI domain. As Mark Zuckerberg put it, while their previous Lama releases have been state-of-the-art relative to open-source models, they want future Lama releases to be state-of-the-art full-stop. Well, they just released a new update to their code generation AI model, Code Lama 70B. With this new announcement, Zuckerberg wrote,

We're open sourcing a new and improved code Lama, including a larger 70 billion parameter model. Writing and editing code has emerged as one of the most important uses of AI models today. The ability to code has also proven to be important for AI models to process information in other domains, more rigorously and logically. I'm proud of the progress here and look forward to including these advances in Lama 3 and future models as well. Now, in terms of avals, CodeLama 70B scored 53 inaccuracy on a human avow benchmark compared to GBT 3.5's 40, but still behind GBT4-76. Still, I think the most relevant line from that Zuckerberg announcement is this one. The ability to code has also proven to be important for AI models to process information in other domains more rigorously and logically. In other words, the reason that

meta is so focused on improving Lama isn't just because they want it to be competitive when it comes to code generation, but because they view it as key to reaching state-of-the-art in general. Now, the information also identifies Microsoft's AI coding tools as a key vector of their battle with Amazon Web Services. The information writes, If you want to understand how Microsoft's embrace of artificial intelligence is driving growth in its business, take a look at GitHub, a business that bought a few years ago that stores code for software developers around the world. It's no secret the developers have flocked to GitHub co-pilot and AI-powered coding assistant Microsoft developed with OpenAI. But Microsoft is making strides towards its bigger goal,

convincing those customers to also rent its Azure Cloud servers a higher margin business than the AI. Now, the example they use is Goldman Sachs. They say that Goldman Sachs has long used a mix of GitHub, GitLab, and other code repositories, but as they've increased spending on GitHub, buying copilot seats for all of its more than 10,000 software developers, the bank has also increased spending on Azure more than 20% in the last half of the year and is on pace to spend more than 10 million annually this year with the service. Now, part of what makes AI so interesting, is that because of its connection to the existing cloud business of companies like Microsoft, AWS, and Google, it means it's immediately financially relevant, not just in a future state.

Indeed, Reuters reports today Microsoft set for AI-powered revenue surge as stock pulls ahead. Microsoft is expected to report a 15.8% jump in quarterly revenue, its best growth in nearly two years, as rising adoption of its products infused with generative AI fuels demand for its cloud services. So there it is, in one sentence, the feedback loop that these big tech companies are using. AI integration into products increases demand for cloud services. It's really hard to overstate how different this makes AI than other speculative technology moments. The way that technology bubbles burst or dissipate, even if the technology is real, is at some point the hype gets way ahead of the potential for revenue generation,

and markets start to back off a little bit and give it some time to catch up. With AI, the financial results are already being seen right now. Or at least they're partially being seen. Another Reuters piece, Alphabet and meta ad sales in Q4, unlikely to reflect Gen.A.I. Investments basically says that when it comes to generative AI inside the ad business, the benefits of all of those integrations may not be showing up quite yet. Now, this is a little bit different because this isn't just a feedback loop that is increasing demand for an existing product. This is a fundamental change to the way that an existing product works.

Basically, the way that generative AI is finding its way into these ad platforms is that it's requiring these companies to have far fewer employees to actually help create ad copy and collateral. A huge amount of that can now be done by AI, which is why you're seeing downsizing in those departments at companies like Google and Facebook. In other words, right now, the quote-unquote benefits of AI for these companies is showing up in cost savings, not so much in revenue increases. The other thing is that the demand for advertising spend tends to be a little bit more dictated by the larger macro environment than it does from any sort of new technology or cost.

When times are bad, advertising goes down. When times are good, advertising goes up. So it's likely to be a little less closely tied to the implementation of AI. Still, by and large, AI is the big market theme again in 2024. MarketWatch writes BlackRock upgrades U.S. stocks to overweight says AI-driven rally could broaden in the next six to 12 months. writes Market Watch, the stock market's tech-driven rally fueled by investor excitement over artificial intelligence should, quote, broaden out as inflation falls further, the Fed starts to cut rates and the market sticks to its rosy macro outlook. Effectively, what BlackRock is seeing is a convergence between the macro environment, where the Fed is not only no longer raising rates, but is actively starting to

think about when it can bring them down again, but that is also coinciding with this larger productivity theme that comes from AI. Effectively, BlackRock thinks that that could be a huge catalyst for continued gains. Now, it is not just big public companies that are getting in on the AI space. There was a ton of big fundraising news over the last day or two. The former CEO of Cloudera has raised $30.5 million for a new AI startup. It's called semaphore.a.i. And appears to be a B-to-B startup that is meant to help companies more easily design AI tools to automate specific tasks inside of them. And then Rebellions Inc. An AI chip company has raised $124 million in Series B funding. writes Yahoo Finance, Rebellions is one of several emergent players trying to break into the market

for semiconductors that train and run AI, competing against fellow startups from California-based Grock Inc to Toronto's Tenth-Dorrent Inc, as well as established leader in Vidia. Finally, one more piece of news from the War Department in AI. According to Reuters, the Pentagon is planning a new AI-based program to estimate prices for critical minerals, such as nickel, cobalt, and more. This was actually announced on a Pentagon website in October, and as Reuters writes, is part of Washington's broader efforts to jumpstart U.S. production of critical minerals used in weapons manufacturing and the energy transition.

Basically, one of the big challenges for America to re-shore this part of the defense infrastructure is access to these chemicals. And so they're trying to smooth out that acquisition process with the help of AI. Not a huge story on the face of it, but shows just how deeply and cross-cuttingly AI is finding its way into not only Washington, D.C. in general, but specifically to the military establishment. However, that is going to do it for today's extended AI breakdown brief. Appreciate your listening or watching, and until next time, peace.