COT News

Will the leading LLM at the beginning of 2026 still be subject to the reversal curse?

Sep 22, 9:36 PMJan 2, 7:59 AM

56%chance

422472

OptionVotes

YES

1221

779

What AI safety incidents will occur in 2025?

Jan 10, 10:21 AMDec 31, 6:29 PM

382227

OptionProbability

Deadly autonomous vehicle accident with driver present

Cybersecurity incident caused by AI written code

Any incident involving prompt injection in production

Another deadly crime planned with the use of LLMs (after Trump hotel explosion)

Any incident that results in internal investigation, caused by the use of AI by law enforcement

Cybersecurity incident caused by AI inference in production

[Duplicate]

Serious incident involving persuasion or blackmail by AI system

Deadly incident involving autonomous manufacturing equipment

Serious incident that can be directly attributed to misaligned agent behavior

Cybersecurity incident that can be directly attributed to misaligned agent behavior

Deadly incident involving autonomous military equipment

Deadly incident caused by AI medical system or equipment

Deadly autonomous vehicle accident with no driver present

100

Best public AI model uses "We" in CoT by EOY 2025?

Jun 1, 7:38 PMDec 1, 11:59 PM

21.83%chance

121345

OptionVotes

YES

1892

528

Will the Northern White Rhinoceros (Ceratotherium simum cottoni) become Extinct by 2100?

Feb 11, 1:07 PMJan 1, 10:59 PM

90.74%chance

81041

OptionVotes

YES

1513

Will the gap between open-weights and frontier models on GPQA Diamond be at most 7%?

Aug 5, 9:17 AMJan 1, 10:59 PM

49.43%chance

8600

OptionVotes

YES

101

Did Google intentionally announce Gemini Ultra in a state barely outperforming GPT-4 to slow the capabilities race?

Dec 8, 2:19 AMJan 1, 4:59 AM

6.52%chance

11496

OptionVotes

YES

528

131

What will be the cool word for "cool" in 2032?

Sep 20, 1:06 AMJan 1, 4:59 AM

18370

OptionProbability

Other

sigma

based

cool

spice

whale

heat

loose

caught

cot

retro

TRANCHE

gas

numb

left

sunk

cought

warm

tepid

luke

damp

jank

sweet

sick

沉着的 chénzhuóde

cloe

swag

Will research demonstrate an advantage to uninterpretable CoT before 2030?

Mar 15, 11:34 AMJan 31, 8:59 PM

74.92%chance

9308

OptionVotes

YES

173

When will a language model be fine-tuned via self-play or expert iteration and achieve significant performance increase?

Dec 17, 4:47 PMJan 1, 4:47 PM

12.92%chance

14272

OptionProbability

YES

1001

976

Will "the case for CoT unfaithfulness is overstated" make the top fifty posts in LessWrong's 2024 Annual Review?

Sep 30, 4:51 PMFeb 1, 12:00 AM

23.09%chance

6141

OptionVotes

YES

1089

591

Will "Optimizing The Final Output Can Obfuscate CoT..." make the top fifty posts in LessWrong's 2025 Annual Review?

Jul 31, 1:32 AMFeb 1, 12:00 AM

14%chance

122

OptionVotes

YES

1000

Non-token based CoTs in SOTA reasoning models EOY 2026?

Jul 21, 5:03 PMDec 31, 11:59 PM

115

OptionProbability

All: All SOTA reasoning models use non-token-based CoT

Some: At least one, but not all, SOTA reasoning models use non-token- based CoT

None: No SOTA reasoning models use non-token-based CoT

COT News

Prediction markets for COT

Prediction markets for COT

Prediction markets for COT

Will the leading LLM at the beginning of 2026 still be subject to the reversal curse?

What AI safety incidents will occur in 2025?

Best public AI model uses "We" in CoT by EOY 2025?

Will the Northern White Rhinoceros (Ceratotherium simum cottoni) become Extinct by 2100?

Will the gap between open-weights and frontier models on GPQA Diamond be at most 7%?

Did Google intentionally announce Gemini Ultra in a state barely outperforming GPT-4 to slow the capabilities race?

What will be the cool word for "cool" in 2032?

Will research demonstrate an advantage to uninterpretable CoT before 2030?

When will a language model be fine-tuned via self-play or expert iteration and achieve significant performance increase?

Will "the case for CoT unfaithfulness is overstated" make the top fifty posts in LessWrong's 2024 Annual Review?

Will "Optimizing The Final Output Can Obfuscate CoT..." make the top fifty posts in LessWrong's 2025 Annual Review?

Non-token based CoTs in SOTA reasoning models EOY 2026?

Latest stories