AI Policy Ramifications Aplenty In That Signed Letter Asking For U.S. Support To Pace The AI Frontier

The Pace Of AI Advances

I’m sure that you gamely know that the pace of AI advancements is frenetic. Almost every day there is a new announcement about some resoundingly breathtaking AI innovation. Whereas this used to be a once-a-year kind of pronouncement, we have shifted to daily occurrences. Anyone who does doomscrolling on their smartphone can observe AI breakthrough announcements that arrive on nearly an hourly or minute-by-minute basis.

The ordinary reaction would be that this is an exciting time to be alive. We are all in the front row when it comes to AI advancing and changing our lives. Imagine that fifty years ago the world at large could only dream of such an amazing pace. And, perhaps fifty years from now, in the future, the whole kit-and-caboodle will have slowed down after we’ve already exhausted all feasible AI innovations (well, some believe there will be even more, due to AI generating discoveries on behalf of humans).

Here’s the problem at hand. The pace of technological advancement is outdoing the pace of figuring out how to handle the ramifications of this newest AI. Policies about guiding AI development and controlling its downsides are slowly being churned out. Laws that protect the public from runaway AI are only now being crafted and potentially put in place. The issue is that the AI tech advances are happening at lightning speeds and we are collectively far beyond the end of our skis. For my detailed coverage of this head-scratching conundrum, see the link here.

What To Do About The Pace

There is a tremendous amount of handwringing going on about the feverish pace of AI. One perspective is that it is impossible to hold back progress. No matter what roadblocks or guideposts are erected, AI is going to be advanced. If one country decides to go slower, all that this will do is put themselves behind the times. They are going to be left in the dust.

An associated argument is made that if there is to be a slowing of the pace, this must be done universally. Everyone must agree with it. In that case, no country will get ahead of anyone else. No country will fall behind anyone else. The world will be at the same juncture when it comes to advancing AI. This is the proverbial kumbaya version of AI pacing.

A big question arises that makes the pacing consideration somewhat muddled. How can we know what the pace of AI is? In other words, suppose that all the countries of the globe agreed to slow down the pace of AI advancements. You cannot just naively assume that this will actually occur. It would require a practical means of detecting AI advancements, along with gauging the pace at which they are proceeding. I suppose you could liken this to nuclear missile agreements, namely that without a means of monitoring and measuring, any agreement limiting their proliferation is a mere paper-based “feel good” exercise.

Devising Tools About Pacing

One thing that the AI field seems to be adept at is developing high-tech tools of all kinds. AI itself is a type of tool. Furthermore, AI is surrounded by a slew of tools that aid in keeping the infrastructure up and running, a topic referred to as harness engineering that I’ve extensively covered at the link here.

I bring this up to point out that the aim to devise tools to measure and monitor the pace of AI advancements is certainly in the realm of possibility. It is not some far-fetched notion. There are already existing tools that can aid in this. In that sense, we are already on the conjectured pathway.

The rub is that these AI pacing tools need to be further improved, and they need to be something that the world at large will be willing to adopt. If the tools are burdensome or costly, that’s going to raise challenges. Another worry would be that any such tool might be a Trojan Horse. A country adopts a pacing tool and suddenly discovers that it is a viral cyberhack. The tool takes down their AI throughout their borders and wreaks havoc across all their nationwide computer systems. Not good.

Pacing The Frontier

You are now primed for the online letter that was posted this week on the topic of AI pacing. According to the posting labeled as “Pacing the Frontier”, the letter is signable by employees of frontier AI companies, as noted this way:

• “If you are an employee at a frontier AI company, you can sign the letter here.”
• “To be verified, you must sign using your corporate email or otherwise provide proof of employment.”
• “This statement was published by employees of frontier AI companies, with organizational support from two independent nonprofits, Guidelight AI Standards and Encode AI.”

That alone already led to some grumbling. Why should only employees of AI frontier companies be able to sign the letter? There are plenty of AI scientists, researchers, developers, and the like who just so happen not to be employees of so-called frontier AI firms. Some feel chagrined that they cannot sign the letter. They are being left out simply due to not working at a said-to-be frontier AI company. For my analysis of what constitutes frontier AI and therefore a frontier AI firm, see the link here.

Another criticism is that an employee of an AI frontier company seemingly only needs to be an employee, in any capacity, even if they have little or no specific role in advancing AI. Should an intern in the mailroom be a signer? Are they on the same level of AI expertise as senior AI developers? And so on.

The counterargument is that such an online posting must have some way of ensuring that the letter will be taken seriously. By having employees of frontier AI companies be the signees, they are the easiest to verify (just inspect their email address or related info), and they seemingly have the nearest seat to what is happening in AI. If the signing was opened to the public at large, this could bring a rush of both suitable signers and crackpot signers. It would be logistically prohibitive to screen them all. A line must be drawn somewhere, and for better or worse, that’s the line that has been drawn in this instance.

The Core Message

The online posting has a short preamble that sets the stage about the importance of AI pacing. After doing so, this is the signatory statement that is essentially being endorsed by the signees:

• “We request that the U.S. government support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development.”

Let’s unpack the sentence.

When trying to get signees for AI-related letters, the wording of a letter can be quite a lightning rod. If the wording seems outsized, the odds are that there might be fewer signees. It is always hard to get agreement on nearly any kind of spirited composition. You will instantly lose people that might have signed had the wording been more measured.

In this case, the letter makes a “request” of the U.S. government. Some might prefer that it say “demand” rather than request. If the wording came out that way, you could easily bet that the controversy would have been through the roof. So, this is a request. Those that are adamant that this is something that must be undertaken by the government would likely feel disappointed that it is only construed as a request. That’s not enough juice.

Support Is The Ask

What is the letter requesting be undertaken by the U.S. government? Aha, the letter asks that the U.S. provide support for an international effort on the AI pacing topic. The effort would be intended to develop technical and governance tools for aiding in pacing frontier AI.

You might be scratching your head on this. The request seems a bit bland. What kind of support and how much support? The letter doesn’t say. In theory, the request could be fulfilled by giving $1 to such an international effort. Or maybe give $100 million. No semblance of magnitude is stated.

Again, this makes signing the letter a lot easier. If the amount was some gargantuan number, the odds are that there might be pushback. The money could be used for something else in the U.S. or the world that needs dire assistance. If the amount was tiny, there might be ridicule that the number is too small and won’t move the needle. It’s a fine line that can quickly cause trouble, no matter what amount might be stated.

Unknowns Are Many

Read that one sentence again and mull it over. What specific or existing international effort is the U.S. to be supportive of? None is specified. Perhaps a new effort is to be established. Maybe a U.N. effort would be utilized. It’s an unknown.

Another aspect that some have wondered about is that the sentence entails supporting the development of AI pacing tools. Imagine that the U.S. did so. The letter doesn’t say anything about the U.S. adopting the use of those tools. It only asks to aid the effort to develop the tools. Suppose the U.S. government jumps in with both feet to support the development but later refuses to adopt the tools. Will the effort of supporting the tool development be seen as wasted effort? Or will at least the existence of the tools be good for their own sake and for others that wish to use the tools?

You might be curious why the U.S. government should be involved in developing the AI pacing tools at all, namely, why not have private enterprise or non-profits develop the tools. Is there a need to bring the U.S. government into it? Some would even slide into worries that the government might try to slow or delay the effort or insert sneaky stuff into the tools. They would claim that it might be better to keep all governments out of the building process. Any government that is involved, not just the U.S. government, might be tempted to tilt the wheel.

The Steps And The First Step

From a systematic perspective, you could assert that these are the six major steps of an AI pacing tool rubric or life cycle:

• (1) Develop and test the needed AI pacing tools.
• (2) Demonstrate that the AI pacing tools are reliable, safe, workable, etc.
• (3) Decide whether the AI pacing tools should be adopted.
• (4) Convince countries and their governments to adopt the AI pacing tools and establish what the AI pacing is going to be.
• (5) Convince AI developers to comply with the AI pacing tools restrictions.
• (6) Enforce international compliance that the AI pacing is being properly adhered to.

That’s quite a mouthful. The sentence in the letter is apparently focused on the first step. There are a lot of colossal steps that follow the first step. How long will this all take? Let’s use the back of a napkin and sketch a timeline for all those steps.

We can be optimistic and say that the first step might take one to two years. This gets us to the year 2028. Assume that it takes one to two more years to undertake steps 2 and 3. That puts us at 2030. Step 4, convincing countries and governments to adopt the AI pacing tools, is a wild guess of how long that would take – maybe 5 years? Perhaps by the year 2035 the AI pacing tools would finally be implemented and underway.

A naysayer would suggest this is probably a dreamy hope. A retort is that you must start somewhere. If we don’t get started, there will be a zero chance of something happening. Every journey begins with a single step, as the adage says.

The World We Live In

Developing the AI pacing tools is only the beginning. Their technical validity, operational security, legal legitimacy, economic consequences, and political acceptability would each become major AI policy debates in their own right. Thus, the letter’s seemingly modest request opens the door to a far larger and more contentious dialogue: not merely whether AI development should be paced, but whether it is even feasible to build trustworthy mechanisms capable of determining how, when, by whom, and under what authority that pacing should occur.

One upbeat angle is that even if there is no action taken in direct response to the letter, at least public awareness has been raised. People will start talking earnestly about the pace of AI. Should the pace continue unabated? Should it be slowed down? Should it be sped up? How can we ascertain what the pace is? Etc.

A final thought for now. The famous visionary Stewart Brand made this insightful remark: “Once a new technology rolls over you, if you’re not part of the steamroller, you’re part of the road.” Some ardently believe that we are going to be pressed into the ground by the rapid advances in AI. The AI steamroller is spiraling out of control. An introspective question is whether humanity can do anything about it or whether we are destined to have AI progress at whatever speed the throttle happens to be on. Think this over. Make sure to stand up and deliberately pace around as you contemplate this heavyweight consideration.