🔭 The Einstein AI model

Thomas Wolf reflects on whether AI will bring a compressed 21st century in this essay about the Einstein AI model.

thomwolf.io/blog/scientific-ai.html
🔭 The Einstein AI model homepage screenshot

About 🔭 The Einstein AI model

This is an essay by Thomas Wolf, published on his blog, that questions whether AI will bring about a "compressed 21st century" in which many scientific discoveries occur within a few years. Wolf responds to the idea in Dario Amodei's essay "Machine of Loving Grace." He argues that, if current trends continue, AI is more likely to produce "a country of yes-men on servers" than a country of Einsteins. He ties this to the difference between answering known questions and asking new ones.

The essay uses a personal account of Wolf's time as a student and researcher, along with historical examples such as Copernicus, special relativity, and CRISPR, to illustrate paradigm-shifting thinking. It also criticizes current AI benchmarks such as "Humanity's Last Exam" and "Frontier Math," which test whether models can find answers to questions that already have known, closed-ended answers. The essay states that the views are the author's opinion.

Key features

  • Critiques the "compressed 21st century" claim from Dario Amodei's essay
  • Personal account of moving from strong student to researcher
  • Historical examples of paradigm shifts, including Copernicus and special relativity
  • Discussion of CRISPR's shift from bacterial immune system to gene editing
  • Argues that current AI benchmarks test answers to known questions
Best for: Readers interested in the debate over whether AI will accelerate scientific discovery, including researchers, students, and people following AI policy and research commentary.
Good for
  • Responds directly to a specific, named essay and quotes its central claim
  • Uses concrete historical examples to support its argument about scientific progress
  • States clearly that the central views are presented as the author's opinion
Watch out for
  • Presents a single author's opinion rather than empirical study results
  • The available text ends mid-discussion of AI benchmarks, so the full argument is not covered

🔭 The Einstein AI model at a glance

Type
Publication
Platforms
Web
Last checked
Oct 6, 2026 · Online when checked
Listed
Oct 5, 2026

Similar publications

Other listings in Articles & Essays with the most in common.

AI Safety Summit Policy Updates AISS 2023

gov.uk/government/topical-events/ai-safety-sum…

The AI Safety Summit 2023 is a major global event that will take place on the 1 and 2 November at Bletchley Park, Buckinghamshire.

Publication

Popular in Articles & Essays

Well-known listings from the same topic.

All 40

🔭 The Einstein AI model FAQ

What is the main argument of the essay?
The author argues that AI is more likely to produce a "country of yes-men on servers" than a compressed 21st century, if current trends continue. He says real scientific breakthroughs depend on asking new questions and challenging accepted ideas.
Does the essay discuss AI benchmarks?
Yes. It says tests such as "Humanity's Last Exam" and "Frontier Math" consist of difficult questions with clear, closed-ended answers, and that they test whether models can find answers to questions already known. The author argues that breakthroughs come from asking new questions instead.
Is something wrong or out of date?

Tell us and an editor will check it.

Own this website?

Claim the listing and suggest updates yourself. An editor checks every claim first.

Claim this listing