GPT-4 #6: Will GPT-4 have at least one qualitatively new capability?
32
51
Ṁ2.2KṀ650
resolved Mar 14
Resolved
YES1D
1W
1M
ALL
Obviously GPT-4 is going to do better on all the major NLP benchmarks. Will it be able to do anything that previous pure LLMs (GPT-3, Gopher, PaLM, etc) couldn't?
This could be either an additional engineered modality (e.g. you can ask it to compose some music and get a sound file out), or some novel capability that I subjectively feel no previous (public) LLM had. For instance: if I can have a conversation with it about a proof I'm working on and the conversation is actually useful in a non-rubber ducking way.
Market resolves N/A if GPT-4 is not released by the close date.
Get Ṁ200 play money
🏅 Top traders
# | Name | Total profit |
---|---|---|
1 | Ṁ34 | |
2 | Ṁ28 | |
3 | Ṁ26 | |
4 | Ṁ23 | |
5 | Ṁ20 |
Sort by:
Related questions
Will Google's Gemini beat GPT4 in terms of capabilities on release?
22% chance
Is GPT-4 (0613) more capable than GPT-4 (0314)?
71% chance
Will GPT-5 have fewer parameters than GPT-4? (1500M subsidy)
15% chance
Will GPT-4 escape?
9% chance
How many parameters will GPT-4 have?
Will GPT-5 be capable of recursive self-improvement?
11% chance
Is GPT4 sentient?
20% chance
Is GPT-4 best? (Thru 2025)
52% chance
GPT-4 #5: Will GPT-4 be a dense model?
2% chance
Could GPT-4 recursively self-improve to AGI with the right cognitive architecture? [@Altimor, twitter]
9% chance