# Topic Review and Next-Writing Plan

**Status: All recommended articles and hubs have been written and integrated.** See the [sub-article index](sub/index.md) for the full set.

This is an editorial planning document for the AI Pedia. It reviews what we currently cover, compares it against recent AI events (through early 2025), and records where to write next.

## 1. Current Coverage Summary

### Starting-point articles
- **History** — strong chronological backbone from antiquity to the generative/agentic era.
- **Hardware & Software** — solid foundations: GPUs, TPUs, neuromorphic chips, frameworks, training, inference, agents, algorithms.
- **Philosophy** — strong on mind–machine, consciousness, alignment, safety, interpretability, labor, surveillance, ethics.
- **Recent Topics** — good on LLMs, multimodal/generative systems, agents, safety/governance, economics, science, phenomena.

### Sub-articles (37 total)
History: preconditions, cybernetics, symbolic AI, expert systems, AI winters, statistical learning, deep-learning revolution, generative/agentic era, transformers.

Hardware/Software: hardware foundations, GPU computing, specialized AI hardware, ML frameworks, training pipelines, inference/serving, agents/tooling, algorithms.

Philosophy: mind–machine, consciousness, knowledge/reasoning, free will/agency, cognitive science, human–AI interaction, labor/economy, surveillance/power, culture/communication, alignment, interpretability, ethics/value pluralism, AI safety/governance.

Recent: large language models, multimodal/generative, agents/autonomy, economic/social impacts, science frontiers, phenomena/public discourse, open questions.

Plus **glossary**, **timeline**, and **sub-article index**.

### Overall assessment
- **Strengths:** history, technical foundations, safety/alignment, LLM/agents, governance.
- **Gaps:** application-domain coverage (education, law, healthcare, finance, democracy), some societal/policy topics (AI welfare/rights, trust/polls, accessibility), and several recent events (public safety debates, model-collapse risk, open-source race).

## 2. Recent AI Events and Phenomena (Last Few Years)

Highlights that should inform next writing:

### 2022–2023
- **ChatGPT launch** (Nov 2022) — mainstream adoption.
- **GPT-4** (Mar 2023) and multimodal systems.
- **Open Letter calling for a 6-month pause** on frontier AI (Mar 2023).
- **Geoffrey Hinton** leaves Google and warns publicly about AI risks.
- **Bletchley Summit** (Nov 2023) — first major international frontier-safety meeting.
- US **Executive Order on AI** (Oct 2023).

### 2024
- **EU AI Act** adopted.
- **Anthropic computer use** and **OpenAI Operator** (2024–2025) — agentic systems.
- **Election-year deepfakes** and disinformation concerns.
- **AI copyright lawsuits** (New York Times vs. OpenAI; authors vs. Meta).
- **California SB 1047** and other state AI-safety proposals.
- **AlphaFold**-style AI in science accelerating.

### 2024–2025
- **Reasoning models** (o1/o3-style) and **chain-of-thought**.
- **Open-source race** — Llama, Mistral, DeepSeek; distillation to small efficient models.
- **Model Context Protocol (MCP)** standardizing agent–tool connections.
- **AI welfare / AI rights** debates gaining attention.
- **Compute governance** and export controls (NVIDIA, China).
- **Energy/water** footprint of data centers becoming public policy.

## 3. Topics Not Yet Covered (or Thin)

Prioritized by relevance and current public debate:

### High priority
1. **AI in public safety debate** — the "pause" letter, Hinton's warnings, and the existential-risk disagreement. We cover safety generally but not the specific events and actors.
2. **AI in elections and democracy** — deepfakes, political ads, AI-generated content, and trust in democratic processes.
3. **AI in education** — tutoring, plagiarism, policy responses, and classroom integration.
4. **AI in law and justice** — legal research, court hallucinations, bias in sentencing, regulation of AI in legal practice.
5. **AI in healthcare** — diagnostics, clinical documentation, medical ethics, and regulatory approval.

### Medium priority
6. **AI and open-source/weights** — benefits, risks, and the frontier-model race.
7. **Model collapse and synthetic-data risk** — training on AI-generated data degrading models.
8. **AI welfare / AI rights** — the emerging debate about machine sentience, suffering, and moral standing.
9. **Public trust and opinion** — polls on AI, acceptance, and fears.
10. **AI and energy/water/climate** — data-center footprint and efficiency.
11. **AI and labor organizing** — writers/actors strikes, automation and unionization.

### Lower priority (still valuable)
12. **AI in finance** — algorithmic trading, fraud, credit, and regulation.
13. **AI and accessibility** — assistive tech for disability and language access.
14. **AI in national security and autonomous weapons** — military applications and arms control.
15. **Emergent abilities and benchmarks saturation** — how capability surprises and evaluation evolve.
16. **AI in mental health and companionship** — therapeutic use and dependency risks.

## 4. Recommended Focus for Next Writing

**First wave (fill the biggest gaps):**
- **AI safety debate & public discourse** — add sub-articles on the 2023 pause letter, Hinton's warnings, and the existential-risk controversy.
- **AI and elections & democracy** — new sub-article under Recent Topics.
- **AI in education** — new application-domain sub-article.
- **AI in law & justice** — new application-domain sub-article.
- **AI in healthcare** — new application-domain sub-article.

**Second wave (deepen existing areas):**
- Expand **Recent Topics** with open-source/models race and model-collapse risk.
- Expand **Philosophy** with AI welfare/rights and public trust/polls.
- Expand **Hardware & Software** with data-center energy/water and compute governance.
- Expand **History** with a 2022–2025 "recent history" sub-article covering the ChatGPT era, safety summits, and regulation.

**Cross-cutting projects:**
- A **"AI and society" hub** linking education, law, healthcare, finance, elections, accessibility, and labor.
- A **"Frontier AI" hub** tying together scaling laws, reasoning models, open-source, compute governance, and safety evaluations.

## 5. Suggested New Articles to Write Next

1. `sub/ai-safety-debate.md` — the 2023 pause letter, Hinton's warnings, and the existential-risk controversy.
2. `sub/ai-elections-democracy.md` — deepfakes, political ads, and democratic trust.
3. `sub/ai-education.md` — tutoring, plagiarism, policy.
4. `sub/ai-law-justice.md` — legal AI, hallucinations, bias.
5. `sub/ai-healthcare.md` — diagnostics, regulation, ethics.
6. `sub/open-source-models.md` — the Llama/Mistral/DeepSeek race.
7. `sub/model-collapse.md` — synthetic-data training risks.
8. `sub/ai-welfare-rights.md` — sentience, suffering, moral standing.
9. `sub/public-trust.md` — opinion polls and acceptance.
10. `sub/energy-water-climate.md` — data-center footprint.

These ten are the strongest candidates because they address prominent recent events and fill clear gaps in application-domain and societal coverage.

---
*This plan should be revisited as the field moves; see [Recent Topics](recent.md) and [sub-article index](sub/index.md).*