So, I spent the weekend reading two papers that Anthropic team released in Sep 2026. The first one on “Detecting misuse of AI” and the second one on “Impact of AI on economic growth & cognitive employment”. In this blog piece, I write on the latter paper, and attach a 2-page summary from Claude of the 57-page paper.  

More interestingly, I sparred with Claude on two areas –

1. How do we extend the Anthropic study  to Indian economy and jobs? and
2. Will China’s robotics and USA’s AI explosion become the Scylla and Charybdis of the world of 2030-2035?

I am enclosing the 57-page Paper and the three Summary Notes, and share the key highlights of the three below:

1. Summary of Anthropic’s 57-page white paper on Impact of AI on USA economy and employment : The core idea is a single model that takes a handful of assumptions about AI (how much of the economy’s cognitive work it can do, how widely it’s adopted, the productivity gain per task, and how much it automates versus assists) and traces out what that implies for GDP, wages, the labor share, and unemployment through 2030 – under three scenarios of Modest, Substantial and Extreme, without attaching probabilities. The headline numbers are stark:
2. Under “modest change,” AI barely moves the needle: GDP is 1.6% above its no-AI path by 2030, unemployment ticks up from 3.8% to 3.9%.
3. Under “extreme change,” AI performs nearly half of today’s cognitive work, GDP is 32% higher, growth hits 15.4% a year, the labor share of income falls from 60% to 45%, and almost one in five cognitive workers is unemployed.
4. The “substantial” scenario sits in between (GDP +8.3%, cognitive unemployment 4.5%) and happens to be very close to what the authors’ own survey of 10,980 US adults implies people actually expect.

• Impact on Indian economy – My Prompt: I was wondering if there could be a similar exercise carried out for an economy like India? The driving factors could be different and additional to the basic macro-economic ones used in this paper. %age of employees in IT services sector, % if employees in GCCs, other knowledge workers, predominance of agricultural workers, construction workers, inexpensive blue collar services etc.  How will developments in AI impact an economy like India?

Response from Claude summarised by Sunil :

India is more complex and we need to analyse outcomes across five segments rather than one economy-wide average – Agriculture, Construction & manufacturing, IT/ITES/GCCs, Other white-collar formal workforce and Informal blue-collar & gig workers; accounting for 48% of the workforce (280mn), 23% , 2% (13-14mn), 8% and 19% respectively. The 2% white-collar segments may get impacted similar to what the Anthropic paper writes. Absolute headcount contraction in the NITI Aayog worst-case range (tech services ~7.5–8M→~6M, customer service ~2–2.5M→~1.8M by 2031), net of continued GCC hiring in more strategic roles – concentrated more in tier-2/3 BPM hubs than in Bengaluru/Hyderabad GCC clusters. We also need to find jobs for the 8-10mn graduates each year.

• Impact of Chinese robotics and USA’s AI together – My Prompt: The Anthropic Paper completely omits the impact of a growing Chinese explosion in robotics, especially humanoid robots doing menial jobs. Will AI growth attack Cognitive employment and Robotics attack non-cognitive jobs?

Response from Claude summarised by Sunil :

Claude thinks that impact of robotics on non-cognitive jobs is not going to happen in the next 10 years, while AI will impact cognitive jobs in this period. So perhaps both impacts many not happen together and the world may get time to adjust to these changes. (I am not so sure, seeing the humanoid robot videos doing household jobs and running 100m in sub-10 secs).