The AI Daily Brief: Artificial Intelligence News and Analysis - NVIDIA Sets Off An AI Stock Frenzy While Elon Warns of the Terminator Scenario

Episode Date: May 25, 2023

Today on The AI Breakdown, NLW reviews comments from Elon Musk and Eric Schmidt at yesterday's WSJ CEO Council Summit about AI risk. He also covers EU and UK AI regulatory efforts. On the Brief, how AI is coming to YouTube shorts, plus NVIDIA sets off an AI stock frenzy.  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 talking about why Elon Musk, Eric Schmidt, and a Turing award winner are sounding the alarm on AI. Before that on the brief, AI comes to YouTube shorts, Nvidia launches an AI stock rally, and much, much more. The AI breakdown is a daily video and podcast all about the most important news and discussions in AI. Please like, subscribe, and share, and learn more at breakdown.network. Welcome back to the AI breakdown brief. all the AI headline news and discussions you need in five minutes or less. Let's kick off with the AI browser wars. AI experiences are, of course, changing how we do everything online,

from searching to researching to getting recipes, to planning our workouts. And part of the battle or part of the arms race is what interfaces we're going to use to actually access these new AI tools. Browsers are one of the major battlefields in that war, and Opera has now thrown their hat into the ring. They've just announced their new AI browser sidekick called Area, which is powered by OpenAI's GPT. It can be used to find info on the web, generate text,

or even provide support for using the browser. Opera tweeted my first day with ARIA, performed real-time search to show open gyms, created a workout plan for my needs, made lazy Twitter mod sexy as hell. Bill Gates recently said that he thought that whoever figures out how to make the best AI personal assistant is going to be the winner of this whole arms race, and ARIA is opera's attempt.

Google's deep mind is bringing together two important trends right now, one short-firm video, and two, AI. The newly consolidated group at Google just tweeted, Our powerful visual language model, Flamingo, is changing the way you can watch YouTube shorts. It automatically generates descriptions for hundreds of millions of videos in their metadata, making them more searchable. YouTube shorts are a massive, massive market. They're viewed more than 50 billion times per day, and it is one of the dominant emerging formats. Flamingo generates descriptions from videos automatically. The value to users is that they're going to be able to much more easily search YouTube shorts than they would have been able to before. For creators, theoretically,

it means that their shorts might be viewed by more people who want to see the type of thing that those videos are actually about. Next up, it is robots, robots everywhere. Another robotics-focused startup figure has raised $70 million. CEO, Brett Adcock writes, announcing figure $70 million series A, from launch to robot in 12 months. One year ago today, we started working on the most ambitious project in my career. Can an autonomous humanoid robot be commercially viable this decade? I was compelled this was possible for the following reasons. Bipetal locomotion and walking has matured, AI algorithms and compute is here,

battery and motor energy and power is sufficient. This seems possible, and it will feel like 50 years of the future was pulled forward. 12 months later, he says they've scaled the team to 50 engineers, deployed Figure 1 their first full-scale humanoid robot, and designed commercial plans for the first robot applications. Figure joins Tesla, who just shared this video of their Optimus robots, as well as 1X's Eve in the robot race. Finally, on today's AI breakdown brief, let's talk about the massive AI stock rally that happened yesterday after Nvidia's first quarter earnings report. The reason the stock market was so excited is that Nvidia massively outperformed analytical.

expectations. Adjusted quarter one revenue came in at 7.19 billion, which was higher than the 6.52 billion that was forecast. But the big one was current quarter revenue, which Invidia estimates to be around 11 billion, which is more than 50% higher than analyst forecasts of 7.15 billion. In video was up 28% on the news, while other AI-related stocks, including C3 and Palantir, were up about 8% each. In total, more than $300 billion was added to the market capitalization of stocks related to AI. As NVIDIA approaches a market capitalization of $1 trillion, it is now the fifth biggest company on Wall Street. Now, Nvidia's incredible results are probably also why there seems to be more and more

attention to competing with the chip manufacturer. Microsoft has its project Athena, which is trying to make custom AI-focused chips. And META also announced a custom AI chip effort recently as well. But for now, NVIDIA rules the roost and there is really no good competitor. And as AI becomes more and more important, they stand to gain even more. That's it for today's AI breakdown brief. If you found it useful, please like, subscribe, and share, and I'll be back soon for the main AI breakdown. Elon Musk, Eric Schmidt, and a Turing Award winner are all sounding the alarm on AI risk.

May 2023 may be remembered in the future as the month the AI safety conversation went mainstream. This started, of course, when former Turing award winner Jeffrey Hinton left Google in part to begin warning about the dangers of AI as he saw them. Specifically, Hinton had grown concerned that the arms race. between companies like his Google and Microsoft was creating such incredible economic pressure to advance at any cost that former principles of safety and ethics and responsibility were being left behind in the race. Now, we've covered Hinton's point a lot on the AI breakdown, but today we have two business leaders who are also making news with their prognostications and concerns about the way

that AI is developing. Both set of comments come from the Wall Street Journal's CEO Council Summit, which happened yesterday. The first set of comments that I'm going to read come from Elon Musk. Now, like him or loathe him, Elon Musk is a headline maker. He is constantly in the media. And so when he discusses different topics, especially tech-related topics, they get a lot of attention. When asked about AI regulation, he said, I've been pushing hard for a long time. You figure out some sort of regulatory body and they start off gaining insight and then I've proposed rulemaking. And then we'll get commented on by the industry. Then hopefully we can have some sort of oversight rules and improve safety just like we do

with aircraft with the FAA and spacecraft and cars with the NHTSA and food and drugs with the Food and Drug Administration. The comments of Elon's that got the most amount of attention were those on AI risk. He said, I don't think that AI is going to try to destroy all humanity, but it might put us under strict controls and there's a non-zero chance of it going Terminator. It's not zero percent, but I think it's a small likelihood of annihilating humanity, but it's not zero. We want that to be as close to zero as possible. And then, like I said, of AI assuming control for the safety of all humans and taking over all the computing systems and weapon systems of Earth and effectively being some sort of Uber nanny. Non-zero chance of going Terminator is just tailor made for headlines, right? When it comes to

more practical concerns, Elon discussed his fears around AI and social media. One of the first places we have to be careful of AI being used is in social media to manipulate public opinion there. And then when asked about the next election, he said, there probably will be attempts to use AI to manipulate the and some of it will be successful, and if not this election, for sure, the next one. It wasn't all doom and gloom, though, when he was asked about AI and society, he said, In terms of access to goods and services, I think AI will be ushering an age of abundance, assuming that we're in a benign AI scenario. I think the AI will be able to make goods and services very inexpensively.

If you say over a 20 or 30 year time frame, I think things will be transformed beyond belief. You probably won't recognize society in 30 years. Finally, on the timeline for AGI, he said, I think we're perhaps only three, maybe six years away from it, this decade. So in fact, arguably, we are on the event horizon of the black hole that is artificial general intelligence. Also speaking at the Wall Street Journal Summit was former Google CEO Eric Schmidt, and he discussed AI as well. Now, Schmidt has been involved in AI for years, but he also has concerns about how it's developing. For Schmidt, it's not about jobs or transformation. It is about existential risk. To put a point on this,

he said, existential risk is defined as many, many, many, many people harmed or killed. Schmidt said, there are scenarios, not today, but reasonably soon, where these systems will be able to find zero-day exploits in cyber issues or discover new kinds of biology. Now, this is fiction today, but its reasoning is likely to be true. And when that happens, we want to be ready to know how to make sure these things are not misused by evil people. Another AI risk warning came this week from Joshua Benjillo.

Benjillo was the 2018 Turing Award winner alongside Jeffrey Hinton, and he just released a long blog post called How Rogue AIs May Rise. Benjio defines rogue AI as an autonomous AI system that could behave in ways that could be catastrophically harmful to a large fraction of humans and potentially endanger societies, species, or the biosphere. Benjillo echoes many of Hinton's concerns with the AI arms race, writing, The competitive nature of capitalism is clearly also a cause for concern, as a potential source. of careless AI design, motivated by profits in winning market share, that could lead to potentially rogue AIs. AI economists may help us one day to design economic systems which rely less on competition and the focus on profit maximization, with sufficient incentives and penalties to counter the advantage of autonomous goal-directed AI that may otherwise push corporations there. The risk of rogue

AIs is scary, but it may also be a powerful motivation to redesign our society in the direction of greater well-being for all. Now, there are plenty of detractors who point out problems with Ben Gio's arguments, but I share it in the context of this broader, increasing conversation about AI safety in the public discourse. This, of course, is entering the political sphere as well. A couple weeks ago, the White House, of course, met with AI CEOs, and now the British government is doing the same, with Prime Minister Rishi Sunnuk hosting OpenAI Sam Altman and others as well to discuss these exact issues. Leaders from Anthropic and DeepMind were also present, and it doesn't really seem like we got a lot of detail about what was discussed except very high-level

trivialities. In the meantime, the EU is advancing what could be pretty onerous AI regulations. OpenAI Sam Altman has in fact said that they might have to leave the continent if the regulations go through as they are currently written. Speaking in London, he said that he had, quote, many concerns about the EU's planned AI Act and that the details really matter. We will try to comply, but if we can't comply, we will cease operating. Here's that the Financial Times describes where we are. The EU's AI Act was initially designed to deal with specific high-risk uses of artificial intelligence, including its use in regulated products, such as medical equipment, or when companies use it in important decisions for granting loans and making hiring decisions.

However, the sensation caused by the launch of ChatGPT late last year has caused a rethink, with the European Parliament this month setting out extra rules for widely used systems that have general applications beyond the cases previously targeted. The latest plan would require makers of foundational models, the large systems that stand behind services such as ChatGPT, to identify and try to reduce risks that their technology could pose in a wide range of settings. The new requirement would make the companies that develop the models, including OpenAI and Google, partly responsible for how their AI systems are used, even if they have no control over the

particular applications the technology has been embedded in. And even as the EU Parliament moves forward with regulation, Google is also meeting with EU ministers to develop a sort of voluntary AI pact in advance of any regulation being brought to bear. EU industry chief Tieri Breton said on Wednesday, Sundar Pichai, the CEO of Google, and I agree that we cannot afford to wait until AI regulation actually becomes applicable and to work together with all AI developers to already develop an AI pact on a voluntary basis ahead of the legal deadline. Now, for my money, the most interesting

conversations about AI risk and AI safety happen, not in the halls of Washington or Brussels, but on the pages of Twitter. G. Fodor, for example, says, I think Eliezer Yudkowski should spend more time thinking and writing his answer to this question. What can governments do not to decelerate AI, but to alter incentives or deployment resources that would counterfactually accelerate us faster towards solving AI alignment. Eliezer responds little or nothing, unless they're willing to look far enough a field to consider options like a crash project in neuroengineering for augmenting human intelligence beyond that of the smartest current people.

Then he and G photo go back and forth a bit, but that's not really the point. The point is that Twitter, in these circles, is where the actual interesting conversations about what can be done are happening, more so than just politicians checking off boxes that need to be checked. Perhaps that's overly cynical and a regulatory regime will be done well when it comes to AI. But for right now, I wouldn't bet on it. Anyways, guys, that's it for today. If you're enjoying the AI breakdown, please like, subscribe and share it. Go subscribe to the podcast or the newsletter. You can find all of the information about those things at breakdown.network. And until next time, peace.