March 31, 2023

Pausing the AI arms race

An open letter published this week calls on companies to take a six-month pause on training AI models more powerful than GPT-4.

Elon Musk, Steve Wozniak and 1,800+ others have signed on, urging that “powerful AI systems should be developed only once we are confident that their effects will be positive and their risks will be manageable.” This would, the letter notes, ensure systems are “more accurate, safe, interpretable, transparent, robust, aligned, trustworthy, and loyal.”

But what would a pause really do? Two of Ballistic Ventures’ General Partners shared their thoughts from a cybersecurity perspective.

On the open letter, Ballistic GP Roger Thornton said:

“It’s a beautiful sentiment, reminiscent of Einstein’s effort to rally scientists around the threats of nuclear weapons over 75 years ago. As a cybersecurity practitioner, wouldn’t it be wonderful if we could limit dangerous new technology?” 

“Frankly, I would start with eliminating texting and driving before AI. The reality is that’s never going to happen, so if some small subset agree to it, wonderful. Merciless competitors and nefarious actors are not going to slow down. All this would mean is that the kind-hearted people who support this would find themselves behind in an AI arms race.”

Ballistic GP Barmak Meftah added:

“In principle, there’s a strong case to be made in support of what they are saying. Without a framework and some rules of engagement, this could potentially spin out of control. Disinformation campaigns, tampering and poisoning the training data sets, and hacking the inference engine algorithms are just some of the unintended consequences of a completely framework-free approach to AI.”

“Having said that though, practically speaking, a six-month pause wouldn’t solve anything material because there’s no real plan in place to begin with. How would you realistically control this? Government regulation? A commercial AI consortium that sets the rules of engagement? It’s hard to imagine how you would practically regulate this. And would our adversaries put a pause on their efforts? Absolutely not.”

Send us your thoughts – just don’t rely too heavily on AI to form opinions for you.