Observed Signal · May 28, 2026 · Analysis · Source: techcrunch · Impact: 3/5 · Sentiment: Neutral
RSI Is the New AGI, Hard to Pin Down
The article examines the rising interest in recursive self-improvement (RSI) — AI systems that iteratively upgrade themselves — and how the term has become a catchall similar to AGI. It surveys recent projects and actors chasing RSI, including Richard Socher’s Recursive Superintelligence, Alex Karpathy’s Auto-Research work, and Adaption’s AutoScientist, while noting examples of AI systems already writing code or winning competitions (Anthropic’s Claude Code, Disarray’s agent). Experts and institutions including Google’s Sundar Pichai and Georgetown’s CSET caution that meaningful, human-free RSI remains speculative and distant, pointing to engineering, compute, verification, and alignment challenges. The piece frames RSI as a contested, imprecise milestone with significant uncertainty about timing and impact despite accelerating agentic research activity.
Discussion of RSI and agentic research signals a potentially important AI-inflection trend with implications for automation, model-driven engineering and productization—relevant to technology and MarTech evolution though still speculative.
Track Anthropic Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Recursive self-improvement (RSI) refers to AI systems that can continuously upgrade themselves with diminishing human involvement.
- Richard Socher launched an entity called Recursive Superintelligence earlier in May 2026 with RSI as an explicit goal.
- Alex Karpathy is developing an Auto-Research project using agent swarms and has made work public via a GitHub repository.
- Adaption, founded by Sara Hooker (Cohere and Google alum), launched a tool called AutoScientist to automate frontier model training.
- Anthropic reported heavy use of its Claude Code tool (a lead programmer estimated nearly 100% of his team’s code was written by the tool) and a Mythos preview survey indicated some engineers believe the system could approach L4 engineer capability.
Connected Companies & Entities
5 Entities mappedOntology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
The Race to Recursive Self-Improving AI
An analysis piece published on 2026-05-26 argues that the AI conversation is shifting from AGI hype toward recursive self-improvement (RSI), which the author views as a likely industry theme by 2027. The article surveys startups and research activity — naming firms such as Anthropic, Recursive Superintelligence, DeepSeek and chip-focused players — and cites financing, alumni networks from DeepMind/OpenAI, and partnerships (e.g., Google/Blackstone) as factors accelerating enterprise AI and prospective RSI efforts. The author discusses potential economic and scientific implications, questions the commercial viability of buzzy RSI startups, and predicts 2027 as the start of a broader “Machine Economy” era driven by self-improving AI and enterprise adoption.
Ramez Naum doubts AI fast takeoff to superintelligence
In this essay, futurist Ramez Naum argues that a rapid 'intelligence explosion' via recursive self-improvement (RSI) is unlikely based on current data. He analyzes OpenAI's and Anthropic's internal reports, which show that AI models significantly boost research productivity but with diminishing returns. Naum estimates that the self-improvement loop is currently 5-10 times too weak to sustain itself, let alone run away. He points to gaps between benchmark performance and real-world research task success, as well as steep scaling costs. While narrow superintelligence in verifiable domains like math and coding is emerging, broad superintelligence remains distant. He calls for better data transparency from AI labs to track progress realistically.
AI containment era: limits to recursive self-improvement
The newsletter assesses the practical limits to recursive self-improvement (RSI) in AI, citing Toby Ord’s paper that generation time and physical constraints (speed of light, Bekenstein bound, Landauer limit) make unbounded RSI unlikely. It highlights growing business adoption of open-weight models — with single-day token-share records at Vercel and companies (Thomson Reuters, Bridgewater, Trainloop) fine-tuning open models to cut costs and improve task-specific performance. New technical releases reshape compute economics: Z.ai released GLM 5.3 and OpenAI published first results for its in-house Jalapeño chip, which reportedly outperforms comparable Nvidia silicon on tokens-per-megawatt. The piece also notes increasing hardware heterogeneity (Cerebras, Fractile/Anthropic deal) and touches on institutional and industry reactions to AI (University of Chicago classroom tech bans; Meta team restructuring discussions).
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.