Will language models solve cryptic crosswords by end of 2026?
Plus
14
Ṁ3292027
73%
chance
1D
1W
1M
ALL
If a large language model can solve >90% of clues from Financial Times cryptic crosswords then yes otherwise no. Crossword layout as context is allowed.
Here's an example FT crossword:

For example, 2 down is "Giant pig, including tail, served up (7)". Pig is "hog", put tail inside to get "htailog" then reverse it to get "goliath".
Get Ṁ1,000 play money
Sort by:
In isolation or given the crossword layout as context?
I would argue that 90% of the challenge is eliminated by having the crossword as context (number of letters, positions of letters). In fact, I wouldn't be surprised if many cryptic crosswords could be solved without the clues - just the crossword itself.
Without the crossword layout, I think there is a high degree of ambiguity and 90% might be impossible.
Related questions
Related questions
Will there be an AI language model that strongly surpasses ChatGPT and other OpenAI models before the end of 2025?
48% chance
Will language models be able to solve simple graphical mazes by the end of 2025?
51% chance
Will a Large Language Model save a human life through medical advice by the end of 2025?
94% chance
Will Transformer based architectures still be SOTA for language modelling by 2026?
79% chance
Will we reverse-engineer a language model into an interpretable (python) program by 2027?
9% chance
Will any 10 trillion+ parameter language model that follows instructions be released to the public before 2026?
38% chance
Will someone solve the Scorpion Ciphers by the end of 2032?
38% chance
Will a Large Language Model be listed as an author on a peer-reviewed paper by the end of 2025?
30% chance
Will all of the publicly accessible parts of heavengames.com/aok.heavengames.com become part of a large language model like Claude or GPT by 2025?
59% chance
By the end of 2026, will we have transparency into any useful internal pattern within a Large Language Model whose semantics would have been unfamiliar to AI and cognitive science in 2006?
26% chance