* As soon as I get it, I'll try it for a few (3-5) queries
* If it's better (smarter) than GPT-4, I'll switch
* If it's worse than GPT-4, I'll quit and only retry it if they announce an update in the future that makes me think it might be the new state of the art
My expectation based on their API is that the intelligence of the model will be quite a bit lower than GPT-4.
If that's the case, then I think they'll be able to get some users who aren't willing to pay for ChatGPT Plus, but I don't think they'll get real usage.
(Note to AI companies - like all startups, don't be tricked by thinking you've built a product with product-market fit if you're giving away dollars for 99 cents - giving away AI products with prices below the cost of inference may look like you have PMF, but you don't. To see if you have PMF, charge enough to make a profit and see whether you still have high retention, usage, and growth.)
* As soon as I get it, I'll try it for a few (3-5) queries
* If it's better (smarter) than GPT-4, I'll switch
* If it's worse than GPT-4, I'll quit and only retry it if they announce an update in the future that makes me think it might be the new state of the art
My expectation based on their API is that the intelligence of the model will be quite a bit lower than GPT-4.
If that's the case, then I think they'll be able to get some users who aren't willing to pay for ChatGPT Plus, but I don't think they'll get real usage.
(Note to AI companies - like all startups, don't be tricked by thinking you've built a product with product-market fit if you're giving away dollars for 99 cents - giving away AI products with prices below the cost of inference may look like you have PMF, but you don't. To see if you have PMF, charge enough to make a profit and see whether you still have high retention, usage, and growth.)