# AI & Machine Learning

Canonical URL: https://shipslides.com/d/technology-ai-and-ml
Raw viewer URL: https://content.shipslides.com/d/technology-ai-and-ml/raw
Category: Technology
Slides: 13
Updated: 2026-05-17T20:49:43.320Z
Tags: technology, and

## Summary

A 70-year arc that stalled twice, then accelerated past most of its critics. Below: dates, names, and the equations that built modern AI. Key sections include: The long path from neuron to network.; 1943–1958: The neuron, formalized.; The two AI winters.; 1986: Backpropagation, popularized.; Convolutions and the GPU.; 2012: AlexNet and the spark.; 2017: Attention is all you need.; Scaling laws.; The modern LLM stack.; Multimodal & tool use..

## Slide Outline

1. The long path from neuron to network.
2. 1943–1958: The neuron, formalized.
3. The two AI winters.
4. 1986: Backpropagation, popularized.
5. Convolutions and the GPU.
6. 2012: AlexNet and the spark.
7. 2017: Attention is all you need.
8. Scaling laws.
9. The modern LLM stack.
10. Multimodal & tool use.
11. Agents.
12. Alignment & safety.
13. Watch this.

## Slide Transcript

### Slide 1: The long path from neuron to network.

- Deck 01 / Modern editorial
- A 70-year arc that stalled twice, then accelerated past most of its critics.
- Below: dates, names, and the equations that built modern AI.
- Figure 1. Procedurally seeded image, picsum.photos. Decorative.

### Slide 2: 1943–1958: The neuron, formalized.

- In 1943, Warren McCulloch and Walter Pitts proposed a binary threshold model of the
- neuron — a logic gate with weighted inputs. Fifteen years later Frank Rosenblatt built the
- Mark I Perceptron at the Cornell Aeronautical Laboratory, a 400-photocell machine that could learn
- to distinguish marked cards. The New York Times announced an "embryo of an electronic computer
- that the Navy expects will be able to walk, talk, see, write, reproduce itself."
- Figure 2. The Rosenblatt perceptron: y = step(Σ wᵢxᵢ + b).

### Slide 3: The two AI winters.

- In 1969 Marvin Minsky and Seymour Papert published Perceptrons, proving the single-layer
- model could not learn XOR. Funding dried up. A second winter followed in the late 1980s and early
- 1990s when expert systems failed to scale economically.
- "There is no reason to suppose that any of these virtues carry over to the many-layered version."
- — Minsky & Papert, Perceptrons, 1969 (later revised)

### Slide 4: 1986: Backpropagation, popularized.

- Rumelhart, Hinton, and Williams' Nature paper "Learning representations by back-propagating errors"
- showed that gradient descent through a chain rule could train multi-layer networks. The math had been
- derived by Seppo Linnainmaa in 1970 and applied to NNs by Werbos in 1974 — but the 1986 paper made it stick.
- # a tiny pure-python sketch
- for epoch in range(N):
- y_hat = forward(x, W)
- loss = mse(y_hat, y)
- grads = backward(loss, W) # chain rule
- W -= lr * grads # gradient descent

### Slide 5: Convolutions and the GPU.

- Yann LeCun's LeNet-5 (1998) read postal codes with convolutional layers — local receptive fields,
- weight sharing, pooling. The technique waited for hardware: in 2009 Raina, Madhavan and Ng showed
- GPUs could train deep networks 70× faster than CPUs.
- Figure 3. The classical convolutional pipeline (LeNet-5 family).

### Slide 6: 2012: AlexNet and the spark.

- Krizhevsky, Sutskever, and Hinton's AlexNet halved the ImageNet top-5 error rate to 15.3%.
- Two NVIDIA GTX 580s, ReLU activations, dropout, and 60M parameters. The result was so far ahead of
- the field that the deep-learning revolution effectively dates from this paper.
- YearTop-5 errorModel
- 201028.2%NEC-UIUC (SIFT + SVM)
- 201125.8%Xerox
- 201215.3%AlexNet
- 20146.7%GoogLeNet
- 20153.6%ResNet-152

### Slide 7: 2017: Attention is all you need.

- Vaswani et al. dropped recurrence entirely. Self-attention computes a weighted average of values,
- with weights from scaled dot-products of queries and keys.
- Attention(Q, K, V) = softmax( Q · Kᵀ / √dₖ ) · V
- Parallelizable across sequence positions, the transformer scaled to GPT-3's 175B parameters by 2020
- and beyond. Every modern frontier model — GPT, Gemini, Claude, Llama — is a transformer or close descendant.

### Slide 8: Scaling laws.

- Kaplan et al. (2020) and Hoffmann et al. (2022, "Chinchilla") found loss falls as a power law in
- parameters, data, and compute. The Chinchilla update: for a fixed compute budget, you want roughly
- equal scaling of parameters and tokens (~20 tokens per parameter).
- Figure 4. Stylized loss vs. compute (Kaplan/Hoffmann scaling).

### Slide 9: The modern LLM stack.

- Pretraining
- Self-supervised next-token prediction on web text, code, books, and licensed corpora. Trillions of tokens.
- SFT
- Supervised fine-tuning on curated demonstrations. Teaches the model the desired output format and tone.
- RLHF / RLAIF
- Reinforcement learning from human or AI preferences. PPO, DPO, or constitutional methods.
- Inference
- KV-cache, speculative decoding, quantization, MoE routing. The serving layer is now a research field of its own.

### Slide 10: Multimodal & tool use.

- CLIP (2021) tied images and text into a shared embedding space. By 2024 frontier models were natively
- multimodal: text in, text-image-audio-video out. Tool use — function calling, browsing, code execution —
- turned chatbots into agents that can act.
- CLIPDALL-ESoraGeminiClaude

### Slide 11: Agents.

- An agent is a model in a loop with tools and memory. The 2025–2026 wave — Claude with computer use,
- OpenAI Operator, Devin, AutoGPT descendants — pushed reliability past the threshold for real work:
- software engineering, research, customer support, ops.
- while not done:
- obs = env.observe()
- thought, action = model(obs, history)
- obs = env.act(action)
- history.append((obs, action))

### Slide 12: Alignment & safety.

- The technical problem: train a system whose behavior matches human intent across distribution shift.
- Key concepts include reward hacking, deceptive alignment, eval-gaming, and scalable oversight. The field
- draws from RL, mechanistic interpretability, and formal verification.
- "The genie does what you ask, not what you want."— folk maxim of the alignment community

### Slide 13: Watch this.

- Watch: transformers explained
- Open problems
- Sample-efficient continual learning without catastrophic forgetting.
- Robust mechanistic interpretability of large transformers.
- Scalable oversight of superhuman models.
- Energy and water cost of inference at planetary scale.


## Related Decks

- [CPUs and Architecture](https://shipslides.com/d/technology-cpus-and-architecture)
- [Artificial Intelligence: From Turing to Transformers](https://shipslides.com/d/technology-artificial-intelligence)
- [Biotechnology](https://shipslides.com/d/technology-biotech)
- [Blockchain Technology](https://shipslides.com/d/technology-blockchain)