The AI Daily Brief: Artificial Intelligence News and Analysis - Will Big Tech's AI Needs Solve Energy Issues?

Starting point is 00:00:00 Today on the AI breakdown, we're talking about the good and the bad of big tech and AI. The AI breakdown is a daily podcast and video about the most important news and discussions in AI. Go to Breakdown.net Network for more information about our YouTube, Discord, and our newsletter. Hello, friends, it is the weekend, and that means it's time for another long reads episode of the AI breakdown. Today, we are going to build this off of two different opeds, both of which are focused on very different topics, but which have an intersection in the role of big tech. The first we're going to read is called the AI Races Generating a Dual Reality, and it's by the FTs John Thornhill.

Starting point is 00:00:46 John writes, Two seemingly contradictory stories are told about the impact of artificial intelligence. The first is that the industry will be dominated by a handful of tech giants which boast the data, compute power, and expertise to transform our lives. These will make the most money. The second is that AI is a wildly disruptive technology that is going to kick over the chessboard on which the current economy is played, enabling nimbler insurgents to invent new games. The reality is, both stories may be simultaneously true.

Starting point is 00:01:11 The publication this week of the Artificial Intelligence Index report, a 500-page pulse check on the global industry from Stanford University, provides ammunition for both arguments. But most striking is the current omnipresence of the big U.S. companies, including Google, meta, and Microsoft, in terms of research, investment, and AI model development. Private sector companies have certainly captured many of the smartest AI researchers. In 2011, about 41% of newly minted AI PhD researchers in the U.S. in Canada stayed in academia, with the same proportion entering industry. By 2022, only 20% remained in academia, with some 70% joining industry. Those researchers have enabled the U.S. to build 61 of the most notable AI models over the past 20 years, compared with 25 in the EU and the UK combined

Starting point is 00:01:50 and 15 in China, according to the report. But the cost of developing those models has skyrocketed. OpenAI spent 78 million on compute power to train its GPD4 model, while Google spent $191 million on Gemini Ultra, the report estimates. Last year, private sector AI investment in the U.S. totaled 67.2 billion, significantly higher than the next two biggest countries, China, 7.8 billion, and the U.K., 3.8 billion. Some people argue that AI will be the new rail tracks or telecommunication networks of the 21st century economy on which everything else will run. If so, the giant U.S. tech companies may be steadily usurping some of the traditional functions of governments, investment firms, and legislators in building and running the infrastructure themselves, while writing and enforcing the

Starting point is 00:02:28 rules. Says Russell Wald, deputy director of the Stanford Institute for Human-centered Artificial intelligence that produce the report, the main takeaway is that industry dominates. We need to find a way that the public sector still has a seat at the table. But while the U.S. tech giants might produce the most powerful AI models, they cannot control all the ways in which they are applied. On that score, there are vast opportunities for other countries and smaller companies to compete. One of the most intriguing aspects of the Stanford report is how surveys of public perception show that people in emerging economies appear more enthusiastic about the possibilities of AI than those in the developed West. More than 70% of Indonesian Thai and Mexican respondents thought that AI would be more than

Starting point is 00:03:03 more beneficial than harmful, according to an Ipsos survey last year, that compares with just 37% in the U.S. and France. A higher proportion of respondents claim to be active daily users of chat GPT in Pakistan, Kenya, India, and Brazil, than in the U.S. or the U.K., according to another survey by the Schwartz-Rizman Institute. China has been quick to apply AI to real-world uses, accounting for 61% of global AI patents compared with 21% in the U.S. It is also accelerating away from the pack when it comes to industrial robots, installing 21% of the global total. Demography plays a big role in shaping attitudes. About 90% of the world's youth live outside the developed West and are keen to engage with the digital

Starting point is 00:03:37 economy, says Payal Aurora, an Indian-born academic and author of the forthcoming book, from pessimism to promise. To many of them, technology looks like opportunity. Aurora told the Mindaroo Center for Technology and Democracy Conference in Cambridge in the UK this week, pessimism is the privilege of those who can afford to live in despair. We need to burst the pessimism bubble. As others at the conference responded, the dominance of USAI companies risks creating new forms of techno-futalism or data colonialism, as happened with social media. Emerging economies will be the rule takers, not rule makers in this new world order and further stripped of sovereignty.

Starting point is 00:04:08 But something that only reflects current reality, AI may give them a chance to rewrite the script. Hello, friends, quick note before we get back to the show, I'm so excited to share that Super Intelligent is now live. Superintelligent is a platform for fast, fun, and super practical, useful AI learning. We have something like 300 video tutorials adding 30 to 50 each week, covering every top in AI you can imagine from LLMs to image generators to case studies, use cases, basically everything that tells you how to use AI and what to use it on in short, fast four to seven

Starting point is 00:04:43 minute tutorial videos, which are paired with step-by-step instructions that help you actually use these tools as well. It's $20 a month for unlimited access, and I would love to see you there. Check it out at B-super.a.i. That's B-supertai. All right, breakers. Consensus 2024 marks the 10th gathering of the biggest event that's devoted to all sides of the crypto, blockchain, and web3 ecosystems. Join pioneering fingers and builders as they delve into the future of defy and explore game-changing tech from AI to ZK proofs and everything in between. The event is three days of jam-packed content, networking, and so much more. Some of the speakers at the event include Chris Dixon, the founder and managing partner at A16Z Crypto, Sergey Nazaroff, the co-founder of Chainlink, Kathy Wood, the CEO of Arc, Hester

Starting point is 00:05:29 Purs, Commissioner of course, from the U.S. SEC, and Tom Emmer, Republican Majority Whip for the U.S. House of Representatives. Visit Consensus 24.coindex.com to learn more and save 15% on registration with the code breakdown. That is 15% on registration with the code breakdown. So super interesting little piece here. I agree with John entirely that the difference in optimism around AI between the developed world and emerging markets is hugely notable. It's something that I think about a lot. And frankly, I'm not exactly sure what combination of reasons produce the situation. How much of it is depressing effects of media in those bigger developed world markets that focus only on the risks and the downsides of AI versus the more structural fact that one of the big

Starting point is 00:06:12 impacts of AI seems to be bringing people's skills up to parity with other people with more experience than them. One of the big impacts of AI could be that international labor markets are better able to compete with their domestic alternatives. If that's the case, that could be both reason for optimism. in markets like Indonesia, Thailand, and Mexico, and reason for skepticism or nervousness in developed markets like the U.S. My gut tells me it's a little bit less sophisticated than that in general, but I don't know, and I think either way it's an interesting fact.

Starting point is 00:06:41 Now, in terms of how much big tech will dominate, this is a very meaningful question. One of the things that has been surprising, of course, about the rise of AI, is the extent to which it has been big tech companies dominating, as opposed to what we've traditionally seen, which is new startups coming in and debuting and premiering new technology, while other older than, companies are slower to compete. There are tons of reasons why it's not exactly playing out like that in AI, with the obvious exception of Open AI and Anthropic being two of the big leaders. Part of the reason, though, of course, is just the incredible capital needs of competing in this

Starting point is 00:07:11 space. The capital needs are so significant, in fact, that even traditional venture capital has been sidelined because it can't keep up with the need. There are some fairly big implications of this, and I think it's not wrong to ask how there is going to be balance between the increased aggregated power of these big tech companies and every other type of institution and sector. It's why I've thought for a while that one of the unexpected consequences of AI might be an increase in state power, as it feels it needs to claim more power in order to counterbalance these big tech companies. But another potential solution is the rise of open source. This week, we caught glimpses of the first GPT4 class open source models in the anticipated forthcoming

Starting point is 00:07:49 Lama 3400B, and it could be that those sort of open source forces put downward power pressure on the other big tech companies. But now let's shift over to another op-ed, again focused on the big tech companies, which is more optimistic about how they might use their power for good. It's by the editorial board at Bloomberg and is called AI is a humongous electricity hog. That's great. They write, next time you ask chat GPT for a lasagna recipe, consider how much computing power you're using. On a typical day, the AI chatbot handles an estimated 195 million queries, consuming enough electricity to supply some 23,000 U.S. households. By 2026, booming AI adoption is expected to help drive a near doubling of data centers global energy use to more than 800

Starting point is 00:08:28 terawatt hours, the annual carbon emission equivalent of about 80 million gasoline-powered cars. Will this voracious energy appetite undermine efforts to combat climate change? To the contrary, it can and should be harnessed to speed the green transition. It's easy to envision how things could go wrong. In the U.S., power-hungry AI applications are already adding to strains on electricity grids and pushing utilities to burn more fossil fuels. In Ireland, a global computing hub data centers are expected to consume nearly a third of all electricity by 2032, queue of vignette of people unwittingly boiling the oceans in pursuit of the

Starting point is 00:08:57 perfect dog portrait. There's also a more positive scenario. The users and owners of these data centers, including Alphabet, Amazon, Meta, and Microsoft are among the world's largest companies, with ample cash, long strategic horizons, and public commitments to the environment, who better to drive some of the tens of trillions of dollars in investment required to build clean generation, enhance power grids, and achieve net zero carbon emissions. To an encouraging extent, it's already happening. Tech companies have long been top buyers of renewable energy, and have lately breathed life into technologies such as hydrogen storage and small modular nuclear reactors, ideal for providing the stable power that data centers require. The more they invest, the more they'll

Starting point is 00:09:30 help such innovations reach economies of scale, lowering the cost of clean energy for everyone. They might also help solve one of the biggest challenges of renewables. Wind and sun are highly variable, requiring a lot of fossil fuel capacity to fill sometimes extreme gaps between supply and demand. With the aid of AI, data centers can help a grid meet peak demand by dialing back non-essential operations or shifting work elsewhere, a technique that Google has pioneered. In doing so, they can reduce emissions and increase the whole system's resilience. What then can policymakers do? The best approach by far would be a tax on carbon emissions. This would encourage investment in clean energy, help displace fossil fuel generation, and induce more innovation. Officials should also remove bureaucratic obstacles to building

Starting point is 00:10:05 much-needed capacity, especially nuclear. Beyond that, the authorities who approve new data centers should be more selective. They should require, for example, that owners pay for transmission infrastructure instead of shifting the cost to other consumers, and invest in added clean energy capacity that can supply the grid when needed. Some of these conditions already apply in places like Ireland and Singapore. They should be standard everywhere, particularly for the data centers that governments use. Finally, the public needs better information. Although many companies have pledged to achieve net zero emissions, disclosure standards are lacking. Exactly how much energy data centers consume is hard to say. If they all reported their true energy mix, power efficiency and capacity to support the grid,

Starting point is 00:10:38 they'd illuminate best practices, enable better planning, and ensure accountability. To be sure, data processing isn't necessarily the biggest challenge of the green transition. By one estimate, it accounted for less than 2% of global electricity demand as of 2022. Growth forecasts often prove wrong. Technology breakthroughs can change the picture. Yet the need for cleaner energy could hardly be clearer. Even if AI proves to be a bubble, let it be a bubble with benefits. Now, obviously, I don't think it's a bubble. But I do think the point that the incredible need for electricity and power of the AI sector is more likely to produce innovations and forces to increase our generative capacity than it is to just suck up and prohibit others from accessing the

Starting point is 00:11:14 same electricity. I think thinking about it like that and entering the policy conversation with that idea in mind, that the demand need can be a source of incredible opportunity is exactly the right way to think about this sector. Anyways, guys, that is going to do it for today's AIA breakdown. Until next time, peace.

The AI Daily Brief: Artificial Intelligence News and Analysis - Will Big Tech's AI Needs Solve Energy Issues?

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.