On: Anthropic’s $5B, 4-year plan to take on OpenAI

[Tepix]

It's a bit early to tell, isn't it?

If we get more unexpected emergent abilities by scaling the model further, things could get very interesting indeed.

[sharemywin → Tepix]

Would you rather? invest in super intelligent AGI or NOT invest in super intelligent AGI. Especially if one of those emergent abilities is deciding your either with me or against me..lol

[KineticLensman → sharemywin]

That would be the AI version of Pascal's Wager [0]

[0] https://en.wikipedia.org/wiki/Pascal%27s_wager

[pharmakom → sharemywin]

It still remains to be seen if LLMs will lead to AGI.

[bradleyjg → Tepix]

There’s needs to be some difficult barriers to entry beyond having the money to spend on training FLOPS in order for a startup to compete.

I have no idea if there are or there aren’t, but that’s the big question.

[dr_dshiv → bradleyjg]

I mean, this is just the beginning. Just wait till we get actual scifi robots in the next year or so.

FWIW, I do find that Claude (Anthropic's GPT) is often better than GPT4 -- and very fast. Entrants can compete on price, safety, quality, etc.

[ethbr0 → dr_dshiv]

And a big moat is going to be safety... and specifically configuration of safety.

Wouldn't be surprised at all if the major API-based vendors start leaning in on making their safety config proprietary.

If a business has already sunk XXXX hours into ensuring a model meets their safety criteria for public-facing use, they'd rather upgrade to a newer model from the same vendor that guarantees portability of that, versus having to reinvest and recertify.

Ergo, the AI PaaS that dominate at the beginning will likely continue to dominate.

[dr_dshiv → ethbr0]

Excellent point.

Fine tuning is at a low point now, but i expect this to create a moat for the same reasons.

[dTal → dr_dshiv]

I find that Claude is more conversational (better fine tuning), but not as smart as even ChatGPT.

Prompt:

  The original titles and release years in the Harry Potter series are:

  Philosopher's Stone (1997)
  Chamber of Secrets (1998)
  Prisoner of Azkaban (1999)
  Goblet of Fire (2000)
  Order of the Phoenix (2003)
  Half-Blood Prince (2005)
  Deathly Hallows (2007)
  
  Given this, generate a new Harry Potter title, using only the words found in the existing titles. Avoid orderings in the original titles. You may add or remove plurals and possessives.

Results:

ChatGPT: Blood Chamber of the Phoenix's Prisoner

Claude-instant: Chamber Prince Half-Blood Phoenix

[kalkin → dTal]

ChatGPT is more comparable to what Quora/Poe calls Claude+ - slower/more expensive/smarter. Claude-instant is closer to GPT-turbo in that tradeoff space.

[dTal → kalkin]

Both bots are free on poe.com, so one is not more expensive than the other.

[dr_dshiv → dTal]

The question for me is whether they understand complex concepts and can apply them in new areas.

So when I’m doing quantum computing work, I go back and forth between Claude and GPT4 and both complement the other very well.

[phillipcarter → dTal]

I find the opposite, claude-instant seems to generally give me better results for my use case. FWIW gpt-3.5-turbo is good too, just not quite as good.

[comboy → dr_dshiv]

Is it possible to test it somewhere?

[dTal → comboy]

https://poe.com/Claude-instant

It also provides a ChatGPT interface, and a number of other models.