3 ms·

I'm reminded of this internal document leak [0] from Google that came out 9 months ago. While some predictions like opensource models having the upper edge aga

I'm reminded of this internal document leak [0] from Google that came out 9 months ago. While some predictions like opensource models having the upper edge against 'closed AI' have not aged rather well, this one seems rather obvious to anyone; "People will not pay for a restricted model when free, unrestricted alternatives are comparable in quality. We should consider where our value add really is." [0] https://www.semianalysis.com/p/google-we-have-no-moat-and-neither https://www.semianalysis.com/p/google-we-have-no-moat-and-ne...

bluecalm 3y ago
>>While some predictions like opensource models having the upper edge against 'closed AI' have not aged rather well I wouldn't say it didn't age well. I think it didn't age yet. It will take a while until cheap computing power is available and a community around one of those projects organises. In a very small world of NN chess engines it also seemed impossible to catch up to Alpha Zero and all the TPUs Google had available to train it and yet here we are with much stronger nets and amazing pool of resources to continue training and experimenting with some people contributing tens of thousands of dollars of computing power.