The looming AI monopolies

From: POLITICO's Digital Future Daily - Thursday Jan 18,2024 09:02 pm
How the next wave of technology is upending the global economy and its power structures
Jan 18, 2024 View in browser
 
POLITICO's Digital Future Daily newsletter logo

By Derek Robertson

President Joe Biden hands Vice President Kamala Harris the pen he used to sign a new executive order regarding artificial intelligence.

President Joe Biden hands Vice President Kamala Harris the pen he used to sign a new executive order regarding artificial intelligence during an event in the White House on Oct. 30, 2023. | Chip Somodevilla/Getty Images

POLITICO Magazine published an op-ed on Monday from Ganesh Sitaraman, a Vanderbilt University legal scholar and longtime aide to Sen. Elizabeth Warren (D-Mass.); and Tejas N. Narechania, a University of California Berkeley law professor. In it, they warned Washington it needs to enforce robust competition rules on the AI industry before it develops similar market concentration enjoyed by today’s tech giants. (The op-ed drew from a more expansive white paper published in October 2023.)

I called the two scholars yesterday to ask them where they think the most vulnerable pressure points might be in this nascent policy field. We discussed the significant amount of entrenchment that already exists with artificial intelligence, the viability of open-source AI, and the promise of an alliance between those on both the right and left eager to regulate the field. The following conversation has been edited and condensed for clarity:

Is there one particular level of the AI stack where you worry that concentration could be most harmful?

Sitaraman: I don't think in terms of prioritizing any of them, because the harms are serious in each of these areas and they are slightly different in each of the areas. In the context of semiconductor manufacturing, the harms are not just about the limited number of manufacturers, but also their geographic location. That's a difference in the kind of harm, because it has national security and resilience implications. Then there are the cloud questions, which don't raise the same national security issues that chips do. Each of these are important, but as we talked about in the paper, some of the same policy tools can address the problems at each of these layers, too.

What do you see as the easiest regulatory lift for those in government hoping to rein in any AI monopoly now?

Narechania: It depends on what regulator we care about. In chips, given the national security concerns about their use, we can see an impact and have seen some moves to fund more chip development and production here in the United States. We're investing in, potentially subsidizing new competitive providers. The fact that we've done it suggests that we could potentially do more and that's a place where there's room for movement. Having said that, there are also clear analogues to some of the other things that we're talking about: For example, the FCC is moving on net neutrality now. Congress has talked about and thought about non-discrimination regimes, and favoritism and self-preferencing in a variety of contexts, and so given the attention that's being paid to those concerns I think there is a possibility to move on on that.

Is open-source AI a meaningful driver of competition?

Sitaraman: I'm skeptical of the ability of open-source to solve the problems of concentration in the AI sector for a few reasons. First, often when people talk about open-source and about “open” or “closed,” in reality what we're dealing with is something more like a spectrum of how open something is. This is not really precise enough for the conversation.

Second, the layers where concentration occurs, like chips and cloud services, are not solved by open-source. If you think that open-source models are going to solve the problems of concentration across the AI stack, that's just not the case.

Third, one of the challenges for open-source firms at the model layer is that they are dependent on the oligopoly at the lower layers of the stack, and what that means is that we may see in the future foreclosure or pressure put on open-source model developers through things like self-preferencing, or other kinds of benefits that a vertically integrated company that provides both cloud services and their own models might engage in. That’s partly the reason why you have nondiscrimination rules and structural separations between these layers in other sectors. The final thing is that it's not obvious that open-source in a lot of other areas has done what people think it might do here. Firefox, while it had a lot of innovations, is not the dominant browser.

Narechania: I have nothing to add to that.

What do you think of the Biden administration’s steps to address competition in the Executive Order on AI?

Sitaraman: I think the EO does address competition in an important way, and there's good evidence that people in the government see this as an important issue. The Office of Management and Budget, after the EO came out, issued another document that specifically asked agencies to consider competition questions, including questions around self-preferencing.

Narechania: One of the things that's really exciting and heartening about the EO is that it expressly acknowledges and addresses the competition concerns. Government agencies are starting to think about their procurement decisions through the lens of competition, like whether they're potentially entrenching a particular player or monopolist in that space when they decide who to buy from. We have to see what the rest of the executive branch does now that they have these marching orders.

Sitaraman: When you look at the Biden administration's approach overall to competition it’s been very positive on these issues, from the appointments of [Assistant Attorney General for Antitrust] Jonathan Kanter and [Federal Trade Commission Chair] Lina Khan and [former Special Assistant to the President for Technology and Competition Policy] Tim Wu, to the work that they've been doing across the board with their Competition Council and executive orders.

How worried are you that this will change if Biden leaves the White House?

Sitaraman: One of the things that I've been encouraged by is that there has been bipartisan interest in these issues. You have some conservatives who have been out very aggressively talking about the dangers of big tech; you have Lindsey Graham and Elizabeth Warren working on a regulating big tech bill and calling for a new commission. You have [Republican Sen. Chuck] Grassley and [Democratic Sen. Amy] Klobuchar on the antitrust bill that they had in the Senate last year. This is a great story and one where we should be encouraged that there are more people interested in these issues, which across the board both Republicans and Democrats for many decades were not interested in. This is a sea change we've seen in the last few years, and I think that's a very positive thing.

 

YOUR GUIDE TO EMPIRE STATE POLITICS: From the newsroom that doesn’t sleep, POLITICO's New York Playbook is the ultimate guide for power players navigating the intricate landscape of Empire State politics. Stay ahead of the curve with the latest and most important stories from Albany, New York City and around the state, with in-depth, original reporting to stay ahead of policy trends and political developments. Subscribe now to keep up with the daily hustle and bustle of NY politics. 

 
 
bro-down at mooch-y gulch

A photograph shows a sign of the World Economic Forum (WEF) in the alpine resort of Davos, on the opening day of the annual meeting in Davos on January 16, 2023. - The world's political and business elites gather for the annual Davos summit to promote "cooperation in a fragmented world", with war in Ukraine, the climate crisis and global trade tensions high on the agenda. (Photo by Fabrice COFFRINI / AFP) (Photo by FABRICE COFFRINI/AFP via Getty Images)

A sign at the World Economic Forum's annual summit in Davos. | AFP via Getty Images

Crypto is continuing its self-declared comeback with plenty of attention and happy talk at Davos.

POLITICO’s Morning Money reported today on some of crypto’s biggest machers showing up in Switzerland with an extra spring in their step, inspired by the Securities and Exchange Commission’s recent approval of Bitcoin ETFs.

“The funeral dirge is over,” pro-crypto financier and ambulatory unit of time Anthony Scaramucci said in an interview at Davos. “This is a Lazarus year.”

Coinbase’s chief policy officer Faryar Shirzad told MM that the exchange had a plethora of meetings scheduled with various heads of state and their deputies, and that he was betting on a better reception than from some quarters in the U.S.: “The governments that we’re interested in talking to, the level of interest on their side is typically quite high,” he said. “It’s not as though the Elizabeth Warren attack on us defines how we engage with most governments.”

the real ai act (almost)

European Union lawmakers agreed to final language — really this time — for their AI Act, according to documents viewed by POLITICO.

POLITICO’s Gian Volpicelli and Antoaneta Roussi reported yesterday on the official language for the law, which was reached in late December following the “political deal” announced at the end of Spain’s presidency of the European Union last month. Most significantly, they write that legislators agreed to significant exceptions to a proposed blanket ban on facial recognition after pushback from law enforcement, and restrictions on “general purpose” AI systems with computing power greater than 10^25 FLOPs (floating point operations per second), something that would only apply to the very few most powerful AI models at the moment.

But… as the text was finalized under the Spanish presidency, Gian and Antoaneta note that it will be “supplemented by interpretative provisions, known as recitals” before the truly final text is unveiled on January 24. Some lawmakers have been particularly incensed by the carveouts for facial recognition, which German lawmaker Svenja Hahn said were “an attack on civil rights” and could enable “irresponsible and disproportionate use of biometric identification technology.”

Tweet of the Day

When you talk to young folks, they think that the belief in imminent AGI was caused by the rise of LLMs. In reality, this belief is axiomatic and long predates LLMs. DeepMind, Vicarious were founded in 2010 based on this belief. OpenAI in 2015. Interest in LLMs only began in 2019.In 2015, when deep learning was in its infancy and LLMs were years away, many folks were just as convinced that AGI was around the corner as they are today. The only thing that changed is that people are now anchoring these beliefs on LLMs, whereas in 2015 they were looking to Deep RL, LSTMs, and Neural Turing Machines.

THE FUTURE IN 5 LINKS

Stay in touch with the whole team: Ben Schreckinger (bschreckinger@politico.com); Derek Robertson (drobertson@politico.com); Mohar Chatterjee (mchatterjee@politico.com); Steve Heuser (sheuser@politico.com); Nate Robson (nrobson@politico.com); Daniella Cheslow (dcheslow@politico.com); and Christine Mui (cmui@politico.com).

If you’ve had this newsletter forwarded to you, you can sign up and read our mission statement at the links provided.

 

JOIN 1/31 FOR A TALK ON THE RACE TO SOLVE ALZHEIMER’S: Breakthrough drugs and treatments are giving new hope for slowing neurodegenerative diseases like Alzheimer’s disease and ALS. But if that progress slows, the societal and economic cost to the U.S. could be high. Join POLITICO, alongside lawmakers, official and experts, on Jan. 31 to discuss a path forward for better collaboration among health systems, industry and government. REGISTER HERE.

 
 
 

Follow us on Twitter

Ben Schreckinger @SchreckReports

Derek Robertson @afternoondelete

Steve Heuser @sfheuser

 

Follow us

Follow us on Facebook Follow us on Twitter Follow us on Instagram Listen on Apple Podcast
 

To change your alert settings, please log in at https://www.politico.com/_login?base=https%3A%2F%2Fwww.politico.com/settings

This email was sent to by: POLITICO, LLC 1000 Wilson Blvd. Arlington, VA, 22209, USA

| Privacy Policy | Terms of Service

More emails from POLITICO's Digital Future Daily

Jan 17,2024 09:02 pm - Wednesday

Davos and the global state of quantum

Jan 16,2024 09:50 pm - Tuesday

One step closer to 'Graphene Valley'

Jan 12,2024 09:02 pm - Friday

5 questions for XRA's Liz Hyman

Jan 11,2024 09:02 pm - Thursday

CES braces for an uneasy 2024

Jan 10,2024 09:06 pm - Wednesday

Washington goes to Vegas

Jan 09,2024 09:01 pm - Tuesday

At CES, AI gets in everybody's head

Jan 08,2024 09:19 pm - Monday

Copyright could shape the global AI map