Question

Will transformer derived architectures still be state of the art for language modeling in December 2025?

7 comments7

Total Forecasters61

Community Prediction85%

Make a Prediction

50%

community: 85%

median 85.0%mean 85.5%

Authors:

Matthew_Barnett

Opened:

Closes:

Scheduled resolution:

AI Training and Compute

AI Technical Benchmarks

Artificial Intelligence

Neuroscience

Rnn (software)

Long short-term memory

Software engineering

Google Scholar

Will transformer models be the state-of-the-art on most natural language processing benchmarks on January 1, 2027?

72%

Will transformer derived architectures accelerate progress in deep learning?

30%

Will high-impact research on reducing the sample complexity of Large Language Model pretraining be forthcoming before 2026?

65%

Comments

? comments

Authors:

Matthew_Barnett

Opened:

Closes:

Scheduled resolution:

AI Training and Compute

AI Technical Benchmarks

Artificial Intelligence

Neuroscience

Rnn (software)

Long short-term memory

Software engineering

Google Scholar

OpenAI is reportedly struggling to improve its next big AI model. It's a warning for the entire AI industry.

Business Insider•Nov 12, 2024

Iterative Intelligence and the Dawn of Learner-Centricity

Psychology Today•Nov 14, 2024

Eric Schmidt says there's 'no evidence' AI scaling laws are stopping — but they will eventually

Business Insider•Nov 15, 2024

Learn more about Metaculus NewsMatch

Will transformer models be the state-of-the-art on most natural language processing benchmarks on January 1, 2027?

72%

Will transformer derived architectures accelerate progress in deep learning?

30%

Will high-impact research on reducing the sample complexity of Large Language Model pretraining be forthcoming before 2026?

65%