weights
Adjustable numerical strengths assigned to each connection in a neural network, analogous to synaptic strengths, that the network tunes during learning.
As defined in 68 | Melanie Mitchell on Artificial Intelligence and the Challenge of Common Sense
Where it comes up · 11
- Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history5 claims
- Andrej Karpathy — AGI is still a decade away3 claims
- Demis Hassabis — Scaling, superhuman AIs, AlphaZero atop LLMs, AlphaFold3 claims
- Jeremiah Lowin: Explaining the New AI Paradigm - [Invest Like the Best, EP.307]3 claims
- Paul Christiano — Preventing an AI takeover2 claims
- 434 - Can We Survive Ai2 claims
- 336 | Anil Ananthaswamy on the Mathematics of Neural Nets and AI2 claims
- 68 | Melanie Mitchell on Artificial Intelligence and the Challenge of Common Sense2 claims
and 3 more
Most important claims · 10 of 25Search all claims for “weights” →
- The perceptron, created in the 1950s by psychologist Frank Rosenblatt as a hardware model of linear-threshold neurons, was one of the first brain-based systems that could learn from examples via an adjustable-weight algorithm, but it had no hidden layer.68 | Melanie Mitchell on Artificial Intelligence and the Challenge of Common Sense
- After a training pass, one can mathematically determine which of the model's billions of numbers would move the output in which direction, so one can nudge weights to make noise slightly closer to the objective, though one cannot hand-design a specific picture such as Spider-Man by tweaking numbers.Jeremiah Lowin: Explaining the New AI Paradigm - [Invest Like the Best, EP.307]
- Using a trained network (e.g., ChatGPT inference) is purely a feedforward process where an input vector propagates layer by layer to an output, whereas training requires the back-and-forth of computing a loss on the output side and backpropagating it to update each layer's weights toward the input side.336 | Anil Ananthaswamy on the Mathematics of Neural Nets and AI
- Responsible scaling policies specify, ahead of time, which threats a lab is concerned about, which capabilities it is measuring, the concrete measurement results that would indicate those threats are real, and the protective actions (e.g. securing weights) it would take in response — committing to pause until it can take those actions if it cannot.Paul Christiano — Preventing an AI takeover
- We are not building animals shaped by evolution with hardware baked in; we are building 'ghosts' or spirits—fully digital entities created by imitating human internet data—so analogies to animal learning are misleading and pre-training is a 'crappy evolution' that gives a starting point of knowledge and intelligence.Andrej Karpathy — AGI is still a decade away
- A model consists of an architecture plus parameters called weights, which may number 10 billion or 100 billion; training starts with random weights that produce noise, then repeatedly nudges the weights based on a score of how good the output is until the outputs become sensible.Jeremiah Lowin: Explaining the New AI Paradigm - [Invest Like the Best, EP.307]
- Modern real-world neural networks contain millions or even billions of weights, making it very difficult to figure out what they are actually doing or why they succeed, unlike symbolic AI where humans explicitly build in interpretable rules.68 | Melanie Mitchell on Artificial Intelligence and the Challenge of Common Sense
- Beyond the risk of deployed models being misused, rogue actors and nation-states have strong incentives to steal model weights and fine-tune them for harmful ends, so all frontier labs must use best-in-class cybersecurity, double layers of protection, hardened sandboxes, and eventually air gaps and secure hardware solutions to secure the weights.Demis Hassabis — Scaling, superhuman AIs, AlphaZero atop LLMs, AlphaFold
- A trained neural network can embody in its weights something very similar to what Kepler figured out and can predict future planetary positions as a time series, but it cannot spit out a symbolic form of the underlying equation; extracting the symbolic law from the network's weights is an unsolved problem.336 | Anil Ananthaswamy on the Mathematics of Neural Nets and AI
- Intelligence evolves only on a narrow line: it must be too important and variable to bake directly into DNA weights yet important enough to learn at all; unpredictable, rapidly changing environments that evolution can't pre-bake incentivize building a lifetime learning algorithm, and humans hit a niche (hands, tool use, externalized digestion) rewarding marginal intelligence with a scalable brain.Andrej Karpathy — AGI is still a decade away