Scores by kind of claim
Each kind of claim is scored on its own, out of 100: the share that held up, with the range it likely falls in. Fewer resolved claims give a wider range.
-
Factual claims
Statements about the present or the past, checked against the record: did they hold up?
81 76 87 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
Insufficient record: 2 predictions resolved, 5 needed for a score.
-
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
65 45 83
All types combined: 80 (likely 75 to 85, high confidence), from 84 claims: 78 factual claims, 6 promises. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Demis Hassabis's claims
42 claims
“we always thought that we bet on generality and learning. So those were always at the core of any technique we would use.”
Demis Hassabis, CEO of Google DeepMind 28 February 2024 Demis Hassabis — Scaling, superhuman AIs, AlphaZero atop LLMs, AlphaFold (Dwarkesh Podcast)
DeepMind has always bet on generality and learning as the core of its techniques, focusing on reinforcement learning, search and deep learning.
Mostly true True in substance, wrong in a detail.
The second judge would not resolve this claim; the verdict shown is the first judge's.
The evidence is the person's or their organisation's own records.
Said for an organisation they lead or work for; it counts on both records.
What would count as true or false
True: DeepMind's founding documents, early papers and public statements since about 2010 consistently show general-purpose learning methods (reinforcement learning, deep learning, search) at the core of its approach, as opposed to hand-crafted or domain-specific systems. Mostly true: this was the dominant emphasis, but with some notable exceptions. Mixed: the record shows a mix in which generality and learning were often, but not consistently, central. Mostly false: generality and learning were only occasionally central. False: the record shows DeepMind did not prioritise generality and learning.
Evidence
- Open-sourcing DeepMind Lab DeepMind (deepmind.google blog) · 3 December 2016 Own record
“DeepMind's scientific mission is to push the boundaries of AI, developing systems that can learn to solve any complex problem without needing to be taught how. To achieve this, we work from the premise that AI needs to be general.”
- Human-level control through deep reinforcement learning - Nature Nature (Springer Nature) · 25 February 2015 Own record
“their applicability has previously been limited to domains in which useful features can be handcrafted, or to domains with fully observed, low-dimensional state spaces.”
- Mastering the game of Go with deep neural networks and tree search - Nature Nature (Springer Nature) · 27 January 2016 Own record
“These deep neural networks are trained by a novel combination of supervised learning from human expert games, and reinforcement learning from games of self-play.”
- From unlikely start-up to major scientific organisation: Entering our tenth year at DeepMind DeepMind (deepmind.google blog) · 5 December 2019 Own record
“Many are substantial and exciting contributions to their domains, such as MuZero, which can not only master chess, Go, Shogi but also extend to Atari, without knowing the rules first.”
Reasoning
DeepMind's own published work points to general learning methods: the 2015 DQN paper sets deep reinforcement learning against methods limited to domains with handcrafted features (E2), and the 2016 AlphaGo paper combines deep neural networks, reinforcement learning and tree search (E3). A 2016 blog post states the premise that 'AI needs to be general' (E1), and the 2019 retrospective highlights MuZero mastering several games without knowing the rules (E4). AlphaGo also relied on supervised learning from human expert games (E3), and the evidence includes no founding-era (circa 2010) documents and no independent sources. So the record supports learning and generality as the dominant emphasis, but it does not show the claimed consistency since about 2010.
Cite this claim
Demis Hassabis, “DeepMind has always bet on generality and learning as the core of its techniques, focusing on reinforcement learning, search and deep learning.” (said 28 February 2024). Verdict: mostly true. Reliability API, claim c_2cf5dcdd6a0f, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_2cf5dcdd6a0f
Claims without a verdict
Is a verdict here wrong, or do you speak for Demis Hassabis? Reply on the claim's own page, with the form at its foot (how replies work). Verdicts are made automatically; how verdicts and scores are made.