The technology's potentially far-reaching impacts have spurred a race to shape its future development. Image: REUTERS/Aly Song
WEF

To fully appreciate AI expectations, look to the trillions being invested

John Letzing
Digital Editor, World Economic Forum

  • The public sector is increasingly getting involved in fostering and financing AI’s future.
  • Saudi Arabia, for instance, may establish a $40 billion AI initiative.
  • Such ‘industrial policy’ is generally applied to a country’s most vital interests.

In the last week of March, these two things happened in the world of artificial intelligence: news dropped of a planned AI supercomputer that would cost more than the annual GDP of Bulgaria, and an inveterate tech CEO publicly poked fun at an AI-powered toothbrush that sells for about $140.

But the AI “hype cycle” alluded to by the CEO doesn’t seem anywhere near its peak yet. And in any case, viewing this as just a typical financial bubble might not be an accurate way to frame things.

If, like another seasoned tech CEO, you think AI may be a more meaningful invention than fire, you’re likely in favor of keeping the funding floodgates open.

Seems you’re in luck.

As of last week, Amazon and Microsoft have reportedly committed at least a combined $15 billion to competing generative AI startups. The CEO of one of those startups may yet try to raise as much as $7 trillion more (that’s not a typo), to make the precious chips needed to train models for AI systems more abundant.

Venture capital investors have lavishly funded a pipeline of additional upstarts; eight of the most prominent were recently valued at an average of 83 times their projected annual revenue in the process.

The public sector is also getting in on the action. Saudi Arabia was recently reported to be forming a $40 billion AI initiative, to invest in everything from chipmaking to data centers. It would be a singular vote of confidence in the technology from one of the world’s biggest sovereign wealth funds.

The insatiable appetite for investing in AI is widely shared. | Image: World Economic Forum
The insatiable appetite for investing in AI is widely shared. | Image: World Economic Forum

The overall amount of money funneled into private investments like venture capital deals may have slipped by 2022, but AI remains a high priority for governments around the world. When that gets translated into targeted legislation and funding, it tends to be dubbed “industrial policy.”

This public money can be spent in the form of a fund like Saudi Arabia’s, or via the $30 billion in subsidies the US government is using to attract makers of AI chips (the EU has a similar, €43 billion chip program at least partly focused on AI).

Spending can also come in the shape of a €540 million supercomputer for training AI models, like the one being financed by the EU, France, and the Netherlands, or the £900 million, UK version intended to help that country build its own “BritGPT.” India’s government has a multi-pronged “AI mission” funded with the equivalent of $1.2 billion. China’s spending on AI is projected to surpass $38 billion by 2027.

In the same way that predictions can shape reality, so can mountains of money.

Now, it’s a question of who will do the shaping.

Trying to (mostly) pick winners

Aggressive industrial policy in China is woven into the fabric of its economy. In other parts of the world, it’s only recently become less of a dirty word.

That’s true even in the US, much to the chagrin of some people. It’s a place where the Horatio Alger myth persists, but it’s also one where directing policy and taxpayer money at specific sectors might now be one of the few things the country’s two main political parties can agree on.

The US has actually relied on industrial policy throughout its history a lot more than one might assume. Its public investment in the essential elements of what became the internet, for example, has probably paid for itself a few times over by now. And people still make daily use of infrastructure built as part of the government-funded New Deal established nearly a century ago. But it hasn’t all been smooth sailing.

“Economists in general don’t like industrial policy because they say, well, markets will figure it out,” the economist Laura D’Andrea Tyson said during an “Industrial Policy 2.0” panel discussion at Davos earlier this year. But, she added, “markets don’t pay attention to national security issues.”

And there’s the rub.

Because the potential impacts of AI are so far-reaching, no one wants to be faced with the grave implications of failing to master it and actively participate in molding its future development.

Some places have more homegrown AI investors and startups and others. | Image: World Economic Forum
Some places have more homegrown AI investors and startups and others. | Image: World Economic Forum

Others on the Davos panel were less upbeat. The most generous thing the economist Adam Posen had to offer about industrial policy: “Sometimes it’s coincided with success, and sometimes not.” Putting up money is fine. But people get uncomfortable with the idea of governments propping up “winners” plugged into domestic politics, and shunning better-qualified “losers” without those connections.

Still, industrial policy likely helped spark the Industrial Revolution – so it might be logical for it to play a bigger role in the Fourth Industrial Revolution. For one thing, it’s a means for countries lacking abundant homegrown venture investors and startups to level the playing field.

In a development that risks inducing symptoms of AI fatigue, those venture investors are now not just heavily backing AI startups, they’re also using AI to decide which startups to back. Another potentially off-putting trend: AI’s insatiable appetite for energy. Not to mention, we’re also literally running out of original content to feed AI systems.

It’s natural to want to poke holes in something suddenly so overwhelming. But it’s also true that the collective hive mind isn’t always great at gauging future value (as a cub reporter I was sent into the streets of New York to ask people if they’d buy then-brand-new shares of Google at their IPO price, which would’ve turned each $1 into $30 over the next 15 years, and nearly everyone said “no.”)

It might all boil down to the nature of expectations. Is AI really key to revolutionizing research and improving general well-being, or merely a means to more efficiently perform menial tasks and run content mills while pocketing a lot of money along the way? If we truly believe it’s the former, clinging to orthodoxy about investment strategy may not be the best way forward.

The Saudi Minister of Industry and Mineral Resources also participated in that Davos panel on industrial policy. He had a succinct summary: it’s simply a way to stimulate “some of the things we want to happen faster.”

More reading on AI and industrial policy

For more context, here are links to further reading from the World Economic Forum’s Strategic Intelligence platform:

  • Industrial policy in Europe may be entering a “second golden age,” as governments’ confidence in the ability of private enterprise to spur new markets wavers, according to this piece – but progress should not come at a steep social cost. (Social Europe)
  • A thing that makes us unique as humans, according to this study, is the ability to perform a new task after receiving verbal instructions just once. That is, used to make us unique – AI-powered robots seem to be capable of that now, too. (Science Daily)
  • “We also didn’t get into this industry without the fundamental belief that the future can be made better.” Two venture investors describe ways to improve the lot of healthcare workers with AI (those workers just have to be okay with the idea of being more “fungible”). (Stanford Social Innovation Review)
  • Industrial policy hasn’t just re-emerged as a viable policy option, it’s also become the subject of a growing amount of research, according to this analysis – which may help improve its success rate in the future. (CEPR)
  • About 95% of the solar panels used in the EU come from China, according to this piece, which argues for a smarter industrial policy strategy to reduce that dependence. (Bruegel)
  • Governments are investing in AI, and they’re also using it. This survey digs into public perception of the use of face-recognition technology in particular. (RAND Corporation)
  • When ChatGPT took the world by storm last year it caught many government officials by surprise, according to the piece, which details how the US government will now require notification whenever when a company starts training high-powered AI algorithms. (Wired)

On the Strategic Intelligence platform, you can find feeds of expert analysis related to Artificial IntelligenceCapital Markets, the Fourth Industrial Revolution and hundreds of additional topics. You’ll need to register to view.

Image: World Economic Forum
Image: World Economic Forum

 

Previously posted at :