Summary: Suppose we have many different AI models, none of which we trust isn’t scheming. Should we deploy multiple copies of our most trusted model, or an ensemble of many different models?

I claim that mixed deployment is better, and offer some recommendations.

1. The case for mixed deployment

In a pure deployment, where we deploy multiple copies of our most trusted model, either all our AIs are scheming, or none are.[1] Whereas in a mixed deployment, there might be some models scheming and some not. A pure deployment has the advantage of maximising the chance that no AI is scheming, but the mixed deployment has the advantage of maximising the chance that some AIs aren't scheming. Which advantage matters more?[2]

The optimal deployment depends on how the probability of catastrophe grows with the proportion of scheming AIs.

• If this function is convex, then most danger comes from [...]

---

Outline:

(00:27) 1. The case for mixed deployment

(01:59) 1.1. Why danger might be convex

(03:45) 1.2. Why danger might be concave

(04:48) 1.3. Why danger might be linear

(05:15) 1.4. My overall assessment

(05:53) 2. Recommendations for mixed deployment

(05:58) 2.1. How to deploy diverse models

(06:47) 2.2. How to reduce correlation between models

(07:40) 2.3. How to prioritise research

The original text contained 4 footnotes which were omitted from this narration.

---

First published:

Source:

---

Narrated by TYPE III AUDIO.