3 ms·

If I were an investor, I'd be pretty concerned with such a high valuation after the o1 release. It's great, no question, but in my usage so far it's a modest st

If I were an investor, I'd be pretty concerned with such a high valuation after the o1 release. It's great, no question, but in my usage so far it's a modest step up from 4o, much smaller than the 3->4 jump. Real world exponential growth is exponential until it's logistic, and this sort of feels like entering that phase of the LLM paradigm. Talking to friends who are very successful, knowledgeable AI researchers in industry and academia, their big takeaway from o1 is that the scaling hypothesis appears to be nearing its end, and that this is probably the motivation for trading additional training-time compute for additional inference-time compute. So where does that leave an investor's calculus? Is there evidence OpenAI can pull another rabbit or two out of its hat and also translate that into a profitable business? Seems like a shaky bet right now.

m3kw9 2y ago
Then you alone will not invest
ilaksh 2y ago
They have evidence that inference time and inference time during training can continue to increase the reasoning abilities. They do not actually need any further technology development to continue to add profitable products. There are numerous ways they can apply their existing models to offer services with various levels of specificity and target markets. Even better, their API offerings can be leveraged by an effectively infinite variety of other types of business, so they don't even need to do the application development themselves and can still profit from those companies using their API.
lumost 2y ago
anecdotally, I'm flipping back and forth between o1 and GPT-4. o1 is mildly better at editing larger code segments. I worked with it to edit a large ~2k line python file in an unusual domain. But o1 is also incredibly verbose. It'll respond with 1-2 pages of text which often contains redundant data. GPT-4o is better in it's descriptions.
iamleppert 2y ago
That's exactly what it is and what has been done. It's just an iterative model with a planning step. In my tests you can get nearly the same results with GPT4 by first asking for a plan, and then asking for each step of the plan within the same context window.

m3kw9 2y ago

Then you alone will not invest

ilaksh 2y ago

They have evidence that inference time and inference time during training can continue to increase the reasoning abilities. They do not actually need any further technology development to continue to add profitable products. There are numerous ways they can apply their existing models to offer services with various levels of specificity and target markets. Even better, their API offerings can be leveraged by an effectively infinite variety of other types of business, so they don't even need to do the application development themselves and can still profit from those companies using their API.

lumost 2y ago

anecdotally, I'm flipping back and forth between o1 and GPT-4. o1 is mildly better at editing larger code segments. I worked with it to edit a large ~2k line python file in an unusual domain. But o1 is also incredibly verbose. It'll respond with 1-2 pages of text which often contains redundant data. GPT-4o is better in it's descriptions.

iamleppert 2y ago

That's exactly what it is and what has been done. It's just an iterative model with a planning step. In my tests you can get nearly the same results with GPT4 by first asking for a plan, and then asking for each step of the plan within the same context window.