> ## Content Index
> Fetch the complete content index at: https://www.antoinebuteau.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Lessons from Denny Zhou
- URL: https://www.antoinebuteau.com/lessons-from-denny-zhou/
- Published: 2026-06-25T15:51:22.000Z
- Updated: 2026-07-18T21:50:36.000Z
- Description: Denny Zhou, a Google DeepMind research scientist and founder of its Reasoning Team, helped author Chain-of-Thought prompting. His work examines how generating intermediate steps can move language models beyond simple text prediction toward structured problem-solving.
- Author: Antoine Buteau
- Tags: Profile, AI & Machine Learning Profiles

![Visual summary of operating lessons from Denny Zhou.](https://www.antoinebuteau.com/content/images/2026/06/lessons-from-denny-zhou-profile-infographic.webp)

## Lessons from Denny Zhou

Denny Zhou is a research scientist and founder of the Reasoning Team at Google DeepMind. He is a primary author of Chain-of-Thought prompting, a technique that forces language models to generate intermediate steps before answering questions. This collection documents his approach to moving machine learning past simple text prediction and into structured problem-solving.

### Part 1: Defining Reasoning in AI

1. **On defining reasoning:** "Reasoning is the ability to generate new knowledge from existing knowledge." — [*Source: \[Stanford CS25 Lecture*](https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR%5FZ27CM&ref=antoinebuteau.com)*\]*
2. **On the parrot analogy:** "An LLM is like a parrot that repeats words without understanding their meaning, merely repeating patterns of language." — [*Source: \[Medium Summary of CS25*](https://medium.com/?ref=antoinebuteau.com)*\]*
3. **On memorization vs. logic:** "Traditional machine learning relies heavily on memorizing labeled data, whereas true reasoning requires applying underlying patterns to novel situations." — [*Source: \[KDD Keynote 2023*](https://kdd.org/kdd2023/?ref=antoinebuteau.com)*\]*
4. **On moving beyond pattern matching:** "We must shift our focus from models that simply match patterns to systems capable of explicit, step-by-step thinking." — [*Source: \[Teach Language Models to Reason Talk*](https://www.youtube.com/watch?v=-Srfmszcz2A&ref=antoinebuteau.com)*\]*
5. **On what reasoning is not:** "Reasoning is not just about retrieving the correct final output; it is about the sequential process used to arrive at that conclusion." — [*Source: \[Stanford CS25 Lecture*](https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR%5FZ27CM&ref=antoinebuteau.com)*\]*
6. **On generating novel solutions:** "If a model understands the base instructions, it can derive the correct output for combinations it has never seen during training." — [*Source: \[Research on Analogical Reasoning*](https://arxiv.org/?ref=antoinebuteau.com)*\]*
7. **On the limits of predictive text:** "While large language models are fundamentally probabilistic next-token predictors, we can elicit sophisticated problem-solving behaviors by changing how they are queried." — [*Source: \[Chain-of-Thought Paper*](https://arxiv.org/abs/2201.11903?ref=antoinebuteau.com)*\]*
8. **On the nature of intermediate tokens:** "Reasoning fundamentally relies on the generation of intermediate tokens to build a bridge between the problem statement and the final answer." — [*Source: \[Stanford CS25 Notes*](https://note.com/?ref=antoinebuteau.com)*\]*
9. **On analogical thinking in AI:** "Models can be taught to identify analogies between different structural problems, allowing them to solve new scenarios without direct training data." — [*Source: \[Teach Language Models to Reason Talk*](https://www.youtube.com/watch?v=-Srfmszcz2A&ref=antoinebuteau.com)*\]*
10. **On human-like cognition:** "While we do not claim models 'think' exactly like humans, prompting them to simulate human reasoning steps drastically improves their accuracy on complex tasks." — [*Source: \[Google DeepMind Research*](https://deepmind.google/?ref=antoinebuteau.com)*\]*

### Part 2: The Chain-of-Thought Breakthrough

1. **On the origin of Chain-of-Thought:** "Providing a few-shot 'chain of thought' as an example allows models to break down complex problems—like arithmetic and symbolic logic—into manageable steps." — [*Source: \[Chain-of-Thought Paper*](https://arxiv.org/abs/2201.11903?ref=antoinebuteau.com)*\]*
2. **On altering the solution space:** "Prompting for step-by-step logic is not just polite instruction; it fundamentally reshapes how the model searches its probability space." — [*Source: \[Stanford CS25 Lecture*](https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR%5FZ27CM&ref=antoinebuteau.com)*\]*
3. **On scaling laws and reasoning:** "Chain-of-thought prompting reveals reasoning capabilities that naturally emerge in large language models at scale." — [*Source: \[Chain-of-Thought Paper*](https://arxiv.org/abs/2201.11903?ref=antoinebuteau.com)*\]*
4. **On standard prompting failures:** "Standard prompting often falls short on tasks requiring complex logic because the model attempts to jump straight to the final answer." — [*Source: \[KDD Keynote 2023*](https://kdd.org/kdd2023/?ref=antoinebuteau.com)*\]*
5. **On performance gains:** "By simply asking the model to 'think step by step', we observed massive improvements in math word problem benchmarks like GSM8K." — [*Source: \[Chain-of-Thought Paper*](https://arxiv.org/abs/2201.11903?ref=antoinebuteau.com)*\]*
6. **On data efficiency:** "Reasoning chains allow us to achieve state-of-the-art performance using only a small fraction of the annotated examples typically required for supervised training." — [*Source: \[Teach Language Models to Reason Talk*](https://www.youtube.com/watch?v=-Srfmszcz2A&ref=antoinebuteau.com)*\]*
7. **On interpreting model outputs:** "The intermediate steps generated by chain-of-thought make the model's behavior more interpretable, allowing researchers to see where the logic failed." — [*Source: \[Chain-of-Thought Paper*](https://arxiv.org/abs/2201.11903?ref=antoinebuteau.com)*\]*
8. **On complex arithmetic:** "Tasks that previously required specialized modules can now be solved directly by language models via explicit reasoning chains." — [*Source: \[Stanford CS25 Lecture*](https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR%5FZ27CM&ref=antoinebuteau.com)*\]*
9. **On the simplicity of CoT:** "The power of chain-of-thought lies in its simplicity—it requires no architectural changes to the underlying model." — [*Source: \[Chain-of-Thought Paper*](https://arxiv.org/abs/2201.11903?ref=antoinebuteau.com)*\]*
10. **On few-shot exemplars:** "The quality of the reasoning examples provided in the prompt directly influences the quality of the model's generated logic." — [*Source: \[Google DeepMind Research*](https://deepmind.google/?ref=antoinebuteau.com)*\]*

### Part 3: Self-Consistency and Reliability

1. **On the premise of Self-Consistency:** "Instead of relying on a single decoding path, we sample multiple diverse reasoning paths and select the most consistent answer." — [*Source: \[Self-Consistency Paper*](https://arxiv.org/abs/2203.11171?ref=antoinebuteau.com)*\]*
2. **On mitigating single-path errors:** "Self-consistency acts as an error-correction mechanism, recognizing that a model might occasionally generate a flawed reasoning chain." — [*Source: \[Self-Consistency Paper*](https://arxiv.org/abs/2203.11171?ref=antoinebuteau.com)*\]*
3. **On majority voting in AI:** "By applying a majority vote over multiple generated solutions, we significantly increase the reliability of the final output." — [*Source: \[Stanford CS25 Lecture*](https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR%5FZ27CM&ref=antoinebuteau.com)*\]*
4. **On confidence and logic:** "A model's confidence in an answer strongly correlates with the frequency of that answer across different sampled reasoning paths." — [*Source: \[Self-Consistency Paper*](https://arxiv.org/abs/2203.11171?ref=antoinebuteau.com)*\]*
5. **On orthogonal improvements:** "Self-consistency improves performance independent of the underlying prompt, acting as an orthogonal boost to chain-of-thought." — [*Source: \[Self-Consistency Paper*](https://arxiv.org/abs/2203.11171?ref=antoinebuteau.com)*\]*
6. **On replacing greedy decoding:** "For tasks requiring logic, greedy decoding is suboptimal; sampling diverse paths yields far better results." — [*Source: \[KDD Keynote 2023*](https://kdd.org/kdd2023/?ref=antoinebuteau.com)*\]*
7. **On temperature scaling:** "Introducing temperature into the sampling process allows the model to explore a wider variety of valid reasoning trajectories." — [*Source: \[Self-Consistency Paper*](https://arxiv.org/abs/2203.11171?ref=antoinebuteau.com)*\]*
8. **On robustness to prompt variations:** "Using self-consistency makes the model's performance less sensitive to the exact phrasing of the prompt exemplars." — [*Source: \[Teach Language Models to Reason Talk*](https://www.youtube.com/watch?v=-Srfmszcz2A&ref=antoinebuteau.com)*\]*
9. **On computational cost vs. accuracy:** "While sampling multiple paths increases compute cost at inference, the dramatic reduction in hallucinated answers justifies the trade-off." — [*Source: \[Stanford CS25 Lecture*](https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR%5FZ27CM&ref=antoinebuteau.com)*\]*
10. **On ensemble thinking:** "Self-consistency mimics a committee of experts, where the most commonly agreed-upon answer is usually correct." — [*Source: \[Self-Consistency Paper*](https://arxiv.org/abs/2203.11171?ref=antoinebuteau.com)*\]*

### Part 4: Problem Decomposition and Least-to-Most Prompting

1. **On compositional generalization:** "Least-to-most prompting enables models to tackle novel, complex problems by breaking them down into simpler, previously solved sub-problems." — [*Source: \[Least-to-Most Prompting Paper*](https://arxiv.org/abs/2205.10625?ref=antoinebuteau.com)*\]*
2. **On easy-to-hard solving:** "We can guide language models to solve hard problems by explicitly asking them to solve easier sub-components first." — [*Source: \[Least-to-Most Prompting Paper*](https://arxiv.org/abs/2205.10625?ref=antoinebuteau.com)*\]*
3. **On length generalization:** "Models often struggle with problems longer than those seen in prompts; least-to-most prompting bridges this gap." — [*Source: \[Least-to-Most Prompting Paper*](https://arxiv.org/abs/2205.10625?ref=antoinebuteau.com)*\]*
4. **On recursive problem solving:** "The answer to a sub-problem is appended to the prompt to help the model solve the next sub-problem in the sequence." — [*Source: \[Stanford CS25 Notes*](https://note.com/?ref=antoinebuteau.com)*\]*
5. **On explicit decomposition:** "Unlike standard chain-of-thought, least-to-most explicitly requires the model to generate the list of sub-questions before answering them." — [*Source: \[Least-to-Most Prompting Paper*](https://arxiv.org/abs/2205.10625?ref=antoinebuteau.com)*\]*
6. **On symbolic manipulation:** "For tasks like concatenating the last letters of words, breaking the list into smaller chunks prevents the model from losing track of its state." — [*Source: \[Teach Language Models to Reason Talk*](https://www.youtube.com/watch?v=-Srfmszcz2A&ref=antoinebuteau.com)*\]*
7. **On managing context windows:** "By isolating sub-problems, we reduce the cognitive load on the model's self-attention mechanism at any given step." — [*Source: \[KDD Keynote 2023*](https://kdd.org/kdd2023/?ref=antoinebuteau.com)*\]*
8. **On dynamic prompting:** "The prompt dynamically expands as the model solves each intermediate step, building a verified context for the final answer." — [*Source: \[Least-to-Most Prompting Paper*](https://arxiv.org/abs/2205.10625?ref=antoinebuteau.com)*\]*
9. **On algorithmic execution:** "Least-to-most prompting forces the language model to act more like an algorithm executing a while-loop." — [*Source: \[Stanford CS25 Lecture*](https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR%5FZ27CM&ref=antoinebuteau.com)*\]*

### Part 5: In-Context Learning and Model Optimization

1. **On zero-shot reasoning:** "Models can sometimes perform reasoning tasks without explicit examples simply by appending 'Let's think step by step' to the query." — [*Source: \[Large Language Models as Optimizers Paper*](https://arxiv.org/abs/2309.03409?ref=antoinebuteau.com)*\]*
2. **On instruction tuning:** "Finetuning models on a diverse collection of tasks expressed as instructions drastically improves their ability to reason on unseen tasks." — [*Source: \[Google DeepMind Research*](https://deepmind.google/?ref=antoinebuteau.com)*\]*
3. **On the FLAN collection:** "The FLAN approach demonstrates that exposure to various reasoning structures during tuning creates a more adaptable model." — [*Source: \[Teach Language Models to Reason Talk*](https://www.youtube.com/watch?v=-Srfmszcz2A&ref=antoinebuteau.com)*\]*
4. **On language models as optimizers:** "We can use an LLM to iteratively evaluate and improve its own generated prompts based on task performance." — [*Source: \[Large Language Models as Optimizers Paper*](https://arxiv.org/abs/2309.03409?ref=antoinebuteau.com)*\]*
5. **On automated prompt engineering:** "Instead of humans guessing the best instructions, we let the language model search for the optimal prompt in natural language." — [*Source: \[Large Language Models as Optimizers Paper*](https://arxiv.org/abs/2309.03409?ref=antoinebuteau.com)*\]*
6. **On meta-prompts:** "By providing a meta-prompt that describes the optimization task, the model learns to propose better instructions for itself." — [*Source: \[Large Language Models as Optimizers Paper*](https://arxiv.org/abs/2309.03409?ref=antoinebuteau.com)*\]*
7. **On gradient-free optimization:** "Prompt optimization with LLMs offers a gradient-free way to improve system performance without updating model weights." — [*Source: \[Stanford CS25 Lecture*](https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR%5FZ27CM&ref=antoinebuteau.com)*\]*
8. **On the mathematical foundations of learning:** "In-context learning works because the model's attention mechanism implicitly performs a form of gradient descent on the provided examples." — [*Source: \[KDD Keynote 2023*](https://kdd.org/kdd2023/?ref=antoinebuteau.com)*\]*
9. **On overcoming prompt brittleness:** "Automated optimization helps identify prompts that are mathematically optimal for the model's specific parameter state." — [*Source: \[Large Language Models as Optimizers Paper*](https://arxiv.org/abs/2309.03409?ref=antoinebuteau.com)*\]*

### Part 6: Decoding Strategies and Intrinsic Capabilities

1. **On intrinsic reasoning paths:** "Reasoning capabilities are often inherent in a model's pre-trained probability space, waiting to be unlocked." — [*Source: \[Chain-of-Thought Reasoning Without Prompting Paper*](https://arxiv.org/abs/2402.10200?ref=antoinebuteau.com)*\]*
2. **On the flaws of greedy decoding:** "Always picking the single most likely next token hides the model's true reasoning capabilities." — [*Source: \[Chain-of-Thought Reasoning Without Prompting Paper*](https://arxiv.org/abs/2402.10200?ref=antoinebuteau.com)*\]*
3. **On exploring alternative tokens:** "By examining the top-k alternative tokens during decoding, we often find natural reasoning chains forming without explicit prompts." — [*Source: \[Chain-of-Thought Reasoning Without Prompting Paper*](https://arxiv.org/abs/2402.10200?ref=antoinebuteau.com)*\]*
4. **On CoT without prompting:** "Altering the decoding process allows models to generate reasoning paths purely from their base training distribution." — [*Source: \[Chain-of-Thought Reasoning Without Prompting Paper*](https://arxiv.org/abs/2402.10200?ref=antoinebuteau.com)*\]*
5. **On the hidden state of LLMs:** "There is a massive discrepancy between what a model knows and what it outputs when forced down a greedy path." — [*Source: \[Stanford CS25 Lecture*](https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR%5FZ27CM&ref=antoinebuteau.com)*\]*
6. **On pre-training data:** "The sheer volume of code and step-by-step logic in pre-training corpora bakes latent reasoning structures into the model's weights." — [*Source: \[Teach Language Models to Reason Talk*](https://www.youtube.com/watch?v=-Srfmszcz2A&ref=antoinebuteau.com)*\]*
7. **On decoding as search:** "We must treat text generation not as a linear sequence, but as a search problem over a vast graph of logical possibilities." — [*Source: \[Stanford CS25 Notes*](https://note.com/?ref=antoinebuteau.com)*\]*
8. **On confidence scores:** "Models assign high confidence to tokens that belong to a logical sequence, even if that sequence isn't the absolute top-1 prediction initially." — [*Source: \[Chain-of-Thought Reasoning Without Prompting Paper*](https://arxiv.org/abs/2402.10200?ref=antoinebuteau.com)*\]*
9. **On reducing prompt engineering:** "As we improve decoding algorithms, the reliance on meticulously crafted prompt engineering will naturally decrease." — [*Source: \[KDD Keynote 2023*](https://kdd.org/kdd2023/?ref=antoinebuteau.com)*\]*

### Part 7: The Limitations of Current Models

1. **On susceptibility to distraction:** "Language models are highly sensitive to irrelevant context, which can easily derail their reasoning processes." — [*Source: \[Google DeepMind Research*](https://deepmind.google/?ref=antoinebuteau.com)*\]*
2. **On the hallucination problem:** "When models fail to reason correctly, they often hallucinate confidently, masking logical errors with fluent text." — [*Source: \[Stanford CS25 Lecture*](https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR%5FZ27CM&ref=antoinebuteau.com)*\]*
3. **On out-of-distribution tasks:** "A model's reasoning breaks down quickly when presented with structural rules that contradict its pre-training." — [*Source: \[Teach Language Models to Reason Talk*](https://www.youtube.com/watch?v=-Srfmszcz2A&ref=antoinebuteau.com)*\]*
4. **On the limits of scale:** "Simply making a model larger does not automatically grant it the ability to solve highly compositional logic puzzles." — [*Source: \[Chain-of-Thought Paper*](https://arxiv.org/abs/2201.11903?ref=antoinebuteau.com)*\]*
5. **On arithmetic failures:** "Models still struggle with exact arithmetic on very large numbers because they operate on sub-word tokens rather than digits." — [*Source: \[Stanford CS25 Notes*](https://note.com/?ref=antoinebuteau.com)*\]*
6. **On logical consistency:** "A model might correctly solve a problem one way, but fail when the same problem is phrased with a negative constraint." — [*Source: \[KDD Keynote 2023*](https://kdd.org/kdd2023/?ref=antoinebuteau.com)*\]*
7. **On evaluating reasoning:** "Current benchmarks are inadequate because they measure the final output rather than the validity of the reasoning chain itself." — [*Source: \[Teach Language Models to Reason Talk*](https://www.youtube.com/watch?v=-Srfmszcz2A&ref=antoinebuteau.com)*\]*
8. **On prompt sensitivity:** "The fact that adding a single space to a prompt can change the output demonstrates how fragile LLM reasoning remains." — [*Source: \[Large Language Models as Optimizers Paper*](https://arxiv.org/abs/2309.03409?ref=antoinebuteau.com)*\]*
9. **On memory constraints:** "Long reasoning chains eventually overwhelm the model's attention span, causing it to forget early assumptions." — [*Source: \[Stanford CS25 Lecture*](https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR%5FZ27CM&ref=antoinebuteau.com)*\]*

### Part 8: The Path Toward Artificial General Intelligence

1. **On the path to AGI:** "Achieving Artificial General Intelligence requires moving beyond models that predict text to systems that actively plan and verify logic." — [*Source: \[Teach Language Models to Reason Talk*](https://www.youtube.com/watch?v=-Srfmszcz2A&ref=antoinebuteau.com)*\]*
2. **On model verification:** "The next frontier is building models that can reliably verify their own reasoning steps before generating the final token." — [*Source: \[Stanford CS25 Lecture*](https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR%5FZ27CM&ref=antoinebuteau.com)*\]*
3. **On system 1 vs. system 2 thinking:** "We are transitioning LLMs from fast, intuitive 'System 1' responses to slow, deliberate 'System 2' thinking." — [*Source: \[KDD Keynote 2023*](https://kdd.org/kdd2023/?ref=antoinebuteau.com)*\]*
4. **On inference scaling:** "Future performance gains will come from scaling compute during inference—allowing the model more time to 'think'—rather than just during training." — [*Source: \[Stanford CS25 Notes*](https://note.com/?ref=antoinebuteau.com)*\]*
5. **On autonomous agents:** "Robust reasoning is the prerequisite for deploying language models as autonomous agents capable of executing multi-day tasks." — [*Source: \[Google DeepMind Research*](https://deepmind.google/?ref=antoinebuteau.com)*\]*
6. **On self-improvement:** "Models will eventually use their reasoning capabilities to generate their own high-quality training data, creating a self-improving loop." — [*Source: \[Large Language Models as Optimizers Paper*](https://arxiv.org/abs/2309.03409?ref=antoinebuteau.com)*\]*
7. **On integrating external tools:** "Teaching models to reason includes teaching them when to pause and query external calculators, search engines, or code interpreters." — [*Source: \[Teach Language Models to Reason Talk*](https://www.youtube.com/watch?v=-Srfmszcz2A&ref=antoinebuteau.com)*\]*
8. **On the definition of success:** "Success is a model that says 'I don't know' when it cannot construct a valid chain of thought." — [*Source: \[Stanford CS25 Lecture*](https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR%5FZ27CM&ref=antoinebuteau.com)*\]*
9. **On the ultimate goal:** "Our goal is not just to build a better chatbot, but to create an automated scientist capable of discovering genuinely new knowledge." — [*Source: \[Google DeepMind Research*](https://deepmind.google/?ref=antoinebuteau.com)*\]*