Jason Wei on 3 Key Ideas in AI in 2025

Navigating the Future of Artificial Intelligence

JW

Jason Wei

Research Scientist at Meta Super Intelligence Labs

Previously at OpenAI (co-creator of Q* and deep research) and Google Brain. Popularized Chain-of-Thought prompting and instruction tuning. Over 90,000 citations.

Executive Summary

This presentation synthesizes Jason Wei's analysis of three fundamental ideas that predict the trajectory and impact of AI development in 2025.

Three Key Ideas in AI in 2025

1

Intelligence as a Commodity

AI capabilities are becoming ubiquitous and inexpensive, with costs driven toward zero through adaptive compute.

2

Verifier's Law

AI progress on any task is directly proportional to how easily a correct solution can be verified.

3

The Jagged Edge of Intelligence

AI capabilities are not uniform and will advance unevenly, creating peaks of superhuman performance and valleys of surprising incompetence.

Intelligence as a Commodity

Key Concept

Intelligence and knowledge are becoming ubiquitous and inexpensive, with their cost and access time being driven toward zero.

Two Stages of AI Progress

Stage 1: Pushing the Frontier

Unlocking a new capability that AI could not previously perform well. Progress is gradual, as seen in the steady improvement on benchmarks like MMLU.

Stage 2: Commoditization

Once an ability is achieved, the cost to access that level of intelligence rapidly decreases. For a given performance level on MMLU, the cost in dollars has fallen each year.

Adaptive Compute

A key driver of commoditization is adaptive compute, which allows for varying the amount of compute based on the task. An easy task can be solved with minimal compute, making intelligence cheaper without constantly needing to scale up model size.

The Evolution of Information Retrieval

Decreasing Time to Access Information

Information Retrieval Task Pre-Internet Era Internet Era Chatbot Era Agents Era
Find the population of Busan in 1983. Hours Minutes Instant Instant
Find how many couples got married in Busan in 1983. Days/Weeks Hours Minutes Minutes
Of the 30 most populated Asian cities in 1983, sort them by the number of marriages that year. Weeks/Months Days Hours Hours

Real-World Example

The OpenAI operator model successfully found the number of marriages in Busan in 1983 by navigating the Korean Kosis database—a task that GPT-3 was unable to perform.

Implications of Intelligence as a Commodity

Key Implications

1

Democratization of Fields

Fields historically gated by access to knowledge, such as coding and personal health ("biohacking"), will become more accessible.

2

Increased Value of Private Information

As public information becomes a free commodity, the relative value of private or insider information (e.g., houses available for sale off-market) will increase significantly.

3

Frictionless Access to Information

The future points toward a "personalized internet" where information is retrieved and presented to the user instantly and without friction.

Verifier's Law: The Power of Measurement

Key Concept

The asymmetry of verification is a concept where it is significantly easier to verify a solution than to generate it. This leads to Verifier's Law: The ability to train AI to solve a task is proportional to how easily verifiable the task is.

Examples of Verification Asymmetry

Task Generation Difficulty Verification Difficulty Asymmetry Example
Sudoku Puzzle Medium to Hard Easy Trivial to check if numbers in each row, column, and box are correct
Code to Run Twitter Very Hard Easy Takes thousands to build, but one can verify by using the website
Writing a Factual Essay Easy Very Hard Easy to generate "feasibly true" claims, but fact-checking is tedious

Five Key Factors of Verifiability

Verifiability is a function of five key factors that determine how easily AI can be trained to solve a task:

1

Objective Truth

There is a clear definition of a good versus a bad response.

2

Speed

Verification is fast.

3

Scalability

It's possible to verify millions of proposed solutions at once.

4

Low Noise

The verification process yields the same result every time.

5

Continuous Reward

The quality of a response can be graded on a spectrum, not just as pass/fail.

AlphaEvolve: Leveraging Verifiability

Case Study

DeepMind's AlphaEvolve is a prime example of leveraging the principle of verifiability. By selecting tasks with high verifiability (e.g., finding the optimal placement of hexagons), the system uses an evolutionary algorithm.

How It Works

  1. Samples many candidate solutions from an LLM
  2. Grades them using the objective verification method
  3. Feeds the best solutions back to the model as inspiration for the next round

This approach sidesteps traditional generalization problems by focusing compute on finding the single best answer to a specific problem where the training and testing domains are identical.

Implications

  • The first tasks to be fully automated will be those that are trivial to verify.
  • A significant emerging opportunity is to develop novel ways to measure things, which can then be optimized by AI.

The Jagged Edge of Intelligence

Key Concept

AI capabilities are not uniform and will not advance monolithically. Instead, progress will be uneven, creating a "jagged edge" with peaks of superhuman performance and valleys of surprising incompetence.

Peaks and Valleys

Peaks

Tasks where AI performs exceptionally well, such as hard math problems and some forms of competitive coding.

Valleys

Tasks where AI struggles with seemingly simple concepts (e.g., historically claiming 9.11 is greater than 9.9) or lacks the data to perform (e.g., speaking rare languages).

Implication for "Fast Takeoff"

A sudden, "fast takeoff" of superintelligence is unlikely because AI self-improvement is not a binary switch but a gradual spectrum of capabilities that will develop unevenly across different tasks.

Heuristics for Predicting AI's Rate of Improvement

The rate of AI improvement is not uniform; it varies per task. Here are the key heuristics for predicting how quickly AI will improve at a given task:

1

Digital Tasks

AI progresses much faster on digital tasks due to near-infinite and rapid iteration speed. Physical tasks involving robotics are slower.

2

Human Difficulty

Tasks that are easier for humans tend to be easier for AI.

3

Data Abundance

AI thrives where data is abundant. Performance on math problems, for instance, correlates directly with the frequency of a language in the training data.

4

Objective Metrics as an Exception

A clear evaluation metric can serve as a reward signal, allowing AI to generate its own synthetic data through reinforcement learning and rapidly solve a benchmark even without pre-existing data.

Predicting AI's Impact on Various Professions

Task Human Difficulty Digital Easy to Get Data Likelihood of AI Success
Translation (Top 50 Languages) Easy Yes Yes Already Done
Debugging Basic Code Medium Yes Yes Done (2023)
Competition Math Hard Yes Yes Done (2024)
Conducting AI Research Hard Yes No Near Future (e.g., 2027)
Making a Movie Very Hard Yes Yes Medium Future (e.g., 2029)
Fixing Your Plumbing Medium No Unsure Unlikely Soon
Hairdressing Medium No Unsure Unlikely Soon

Conclusion: The Future of AI in 2025

Key Takeaways

1

Intelligence as a Commodity

As AI capabilities become ubiquitous and inexpensive, we'll see democratization of knowledge-based fields and increased value of private information.

2

Verifier's Law

Tasks that are easily verifiable will be the first to be automated, creating opportunities for businesses that develop novel measurement methods.

3

The Jagged Edge

AI progress will be uneven, with digital, data-rich, and human-easy tasks advancing fastest, while physical and nuanced human-interaction tasks remain largely untouched.

Final Thought

The future of AI is not a uniform "takeoff" but a complex landscape of rapid advancement in some areas and slow progress in others. Understanding these three key ideas provides a framework for navigating the evolving AI ecosystem in 2025 and beyond.

1 / 12