Anthropic research docs
See (vault internal not on exobrain) Anthropic AI impact job. Sticking on here so linux-Claude can see it.
- How Australia Uses Claude: Findings from the Anthropic Economic Index
- Anthropic Economic Index Report: Economic Primitives
- Anthropic Economic Index Report: Uneven Geographic and Enterprise AI Adoption
- Estimating AI productivity gains from Claude conversations
- The Anthropic Economic Index
Let's start with the economic primitives one (it's used in the Oz one).
- "New metrics of AI usage" - primitives
- 5 dimensions:
- user and AI skills
- how complex tasks are
- the degree of autonomy afforded to Claude
- how successful Claude is
- whether Claude is used for personal, educational, or work purposes
- "Augmentation patterns (conversations where the user learns, iterates on a task, or gets feedback from Claude) edged to just over half of conversations on Claude.ai. In contrast, automated use remains dominant in 1P API traffic, reflecting its programmatic nature."
- So how does that translate into actual productivity changes?
- "While substantial concentration remains, since our last report Claude usage has become noticeably more evenly distributed across US states. If sustained, usage per capita would be equalized across the country in 2-5 years."
- That's interesting. So what sectors is it diffusing into and how?
- "Claude succeeds on most tasks, but less so on the most complex ones"
- It's not clear to me here if they mean Claude Code...? Not sure they do.
- Job exposure - depends on the task type. And on education level - "But this simple task displacement would not affect white-collar workers uniformly—for some occupations it removes the most skill-intensive tasks, for others the least."
- And nothing yet on designed around people and how they respond; can we get very different outcomes if different environment? (Cf. Taylorism.)
- "Knowing the success rate of tasks gives a more accurate picture of which tasks might be automated, how impacted certain jobs might be, and how labor productivity will change. Measuring differential performance by user education sheds light on inequality effects."
- What do I think about that? It's excellent data, clearly. I could do with VIEWS on this.
- "Indeed, the close relationship between education levels in inputs and outputs signals that countries with higher educational attainment may be better positioned to benefit from AI, independent of adoption rates alone."
- That's a very interesting point re. how one models tech change and its cohort impact. This was always true...? That's the question isn't it?
- Side-question: does the shipping container count as 'worker tech'? It has Venn overlap certainly.
- O*NET tasks... how to get under the skin, away from atomic task defs? (Not get away from... deepen)
- Regional concentration is roughly a location quotient: ratio of more/less