3 ms·
Hard disagree. My guess is that AI is actually a race-to-the-bottom dynamic. Given the competition across all/every FAANG and tons of startups, my guess is we’l
by aabhay 3y ago
Hard disagree. My guess is that AI is actually a race-to-the-bottom dynamic. Given the competition across all/every FAANG and tons of startups, my guess is we’ll have a wide range of options for APIs across clouds and providers. On the consumer side we’ll have a range of options for chatbots, API integrations, and more.
For most use cases of AI, there is a ceiling to how intelligent it needs to be. I am guessing we’ll be selecting from dozens of models based on various sizes, context lengths, etc. Just like we right-size VMs in the cloud.
- vagab0nd 3y ago"Race-to-the-bottom" might lead to "winner-take-all"? Eventually the profit margin is so thin that only the biggest companies can survive. In other words, once you have a "race-to-the-bottom" situation, it's hard for newcomers to get in the game.
- waveBidder 3y agoml is entirely dependent on your access to relevant data, which itself has strong network effects.
- aabhay 3y agoData access has weaker network effects than you would expect. Generated chats / outputs are rarely good enough as training data, the "best"/"cleanest" data is still expert-created.