6 ms·
For dual-use tech, its relatively common to have a small "government" division alongside the much larger commercial division. If the only difference for governm
by lumost 3y ago
For dual-use tech, its relatively common to have a small "government" division alongside the much larger commercial division. If the only difference for government is fine tuning data/DoD servers - it shouldn't cost OpenAI much to support.
- maxglute 3y agoPeople are escalating AI to strategic/end of world tier of dual use tech. Questionable if it will be treated the same way. OpenAI is not lockheed, but question is how much they value Pentagon contracts or really gov connections especially with all the regulation talk.
- lumost 3y agoThere will certainly be DARPA and other grants to test if dedicated defense suppliers make sense. In some areas, the funding model may be easier as a pure defense initiative where the DoD could shell out 50-100 Billion for cross-service AI functionality. In theory there is no reason why an individual soldier shouldn't be able to order an air strike, or be informed about inbound strikes on their position and how to react to those. Today, this is not possible due to the complex decisioning process for whether a target is worth a strike by X munition vs. Y munition and when. AI could simplify this process. On the other hand, well funded open source and commercial companies may deliver faster and with higher quality. The DoD doesn't use bespoke databases, and microchips for a reason (except when they do).