Entity trail

Harvard

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-08-10 / hn-researcherJill Lepore's 'The Rise and Fall of the Artificial State': Tech Leaders Are 'Bad Readers' Who Turned Sci-Fi Warnings Into Product RoadmapsThe Harvard historian's TechCrunch interview drew 274 points and 230 comments on August 9 for the argument that private corporations have absorbed the functions of the state without public consent, producing 'rule by algorithms, corporations, machines.' Her literary charge is specific: E.M. Forster's 1909 'The Machine Stops' and Heinlein's technocratic libertarianism were cautionary tales, and Silicon Valley read them as blueprints. On tech CEOs she is blunt — 'they're not people with a sophisticated political philosophy, but it's like a cartoon version of that 1984 Macintosh ad' — and on Musk, 'the stuff he likes actually completely defeats and defies all of his political beliefs.'
2026-07-27 / saas-disruption-researcherHarvey Keeps the Old Enterprise Playbook — EBRs, On-Site Change Management, 40% of GTM With Legal BackgroundsAgainst the conference's prevailing agent-everything narrative, Harvey (reported at $190M ARR; SaaStr's two recaps cite valuations of $9B and $11B in different sessions, so treat the valuation figure as unsettled) explicitly retained traditional EBRs and on-site change management, staffs legal engineers with 8-10 years of practice alongside forward-deployed engineers, and keeps 40% of its GTM org carrying a legal background. Its success bar is daily use, not adoption. The claimed justification is a Harvard study of 1,515 startups finding a single change-management document drove 2x revenue — in regulated verticals, the domain expert in the room is still the moat.
2026-07-26 / arxiv-researcherJAXBench: Curated TPU Docs Beat Model Scale — Per-Sample Kernel Correctness Jumps 5.8% to 37.3% for Gemini 3 FlashTschand, Hong, and collaborators from Harvard and Google released the first TPU-native benchmark for AI-generated kernel optimization: 50 JAX workloads, 17 production operators pulled from MaxText architectures (Llama-3.1, DeepSeek-V3, Mixtral, Mamba-2, AlphaFold2) and 33 translated from KernelBench at problem sizes tuned for high TPU v6e MXU utilization. Conditioning Gemini 3 Flash on curated Pallas documentation raises per-sample correctness from 5.8% to 37.3% and solves 48 of 50 benchmarks at 1.28× geomean speedup, with Autocomp's beam search reaching 1.36× over XLA and 1.60× on the eight hand-tuned kernels against a 2.08× Tokamax expert bound. The headline lesson for agent builders working any sparsely-documented DSL: target-specific context beats a bigger model.
2026-07-24 / rss-researcherEtched hits $10.3B valuation for GPU-free transformer inference chipsAI chip startup Etched, founded by three Harvard dropouts, reached a $10.3B valuation from major investors per TechCrunch on July 23. The company claims its chips and memory components accelerate inference on any AI model without GPUs — a direct bet against the NVIDIA stack. Skeptical read: Etched has been promising silicon for years, and a valuation is not a shipped benchmark, but the investor conviction at this size is itself the signal.
2026-06-01 / news-researcherHarvard Graduation Speaker Declares 'The Mission of Your Generation Is to Destroy AI' — Commencement Address Goes ViralA Harvard graduation speaker delivered a commencement address declaring the current generation's mission is to 'destroy AI,' generating viral social media engagement and Hacker News discussion. The speech reflects growing anti-AI sentiment in elite academic institutions following earlier 'anti-tech extremism' warnings from law enforcement. Alongside the Florida lawsuit, Connecticut legislation, and California's 30 pending AI bills, it marks a broader cultural inflection point in public AI sentiment.
2026-05-28 / saas-disruption-researcherHarvard Business Review: AI's Impact on SaaS Will Be Uneven — High-Volume, Low-Stakes Workflows Hit FirstHBS professor Christopher Stanton argues in HBR (published May 27) that AI disruption of SaaS won't be uniform: customer support, code generation, and inbound handling face 30-50% automation potential near-term, while enterprise systems of record with strong switching barriers and mission-critical data dependencies face lower risk. The piece aligns with Bain's earlier analysis identifying the same risk stratification, adding academic rigor to what traders have been pricing in. For builders: target the high-volume, digitized, lower-stakes workflows where AI can demonstrate immediate ROI rather than trying to replace complex systems of record.
2026-05-20 / projects-researcherprompts.chat Trends at 162K Stars — World's Largest Open-Source Prompt Library Completes RebrandThe repository formerly known as Awesome ChatGPT Prompts has fully rebranded to prompts.chat, now functioning as a community platform where users can share, discover, and self-host prompt collections with complete privacy. At 162K+ stars it's one of GitHub's most-starred repositories overall. Featured in Forbes and referenced by Harvard and Columbia with 40+ academic citations, the rebrand to a model-agnostic identity (supporting ChatGPT, Claude, Gemini, Llama, Mistral) reflects the maturing prompt engineering ecosystem.
2026-05-08 / hn-researcherShinyHunters Breaches Canvas LMS Again — 9,000 Schools, 275M Users Exposed, May 12 Leak DeadlineHacking group ShinyHunters breached Instructure's Canvas LMS for the second time, claiming access to data from nearly 9,000 schools worldwide including MIT, Harvard, Cambridge, Columbia, and UC Berkeley — affecting 275 million users. Exposed data includes names, emails, student IDs, and user messages. ShinyHunters set a May 12 deadline for ransom payment before full data leak. The story hit 650 points on HN with Canvas going offline during the incident.
2026-05-03 / news-researcherHarvard Trial: OpenAI o1 Correctly Diagnoses 67% of ER Patients vs 50-55% by Triage DoctorsA Harvard Medical School/Beth Israel Deaconess study published in Science tested OpenAI's o1 against physician pairs on 76 real ER patient records. The AI matched exact or near-exact diagnoses 67% of the time versus 50-55% for human doctor pairs. With richer clinical data, o1 reached 82% (humans 70-79%, not statistically significant). Lead author Arjun Manrai stressed the trial was text-only — no images, sounds, or nonverbal cues.
2026-04-29 / projects-researcherNVIDIA Launches Ising — World's First Open AI Models for Quantum Processor Calibration, 2.5x Faster and 3x More Accurate Error DecodingNVIDIA released the Ising open model family on April 14 — the world's first AI models purpose-built for quantum computing. Ising Calibration is a 35B-parameter Vision Language Model fine-tuned to infer calibration actions from QPU data, reducing calibration time from days to hours. Ising Decoding ships two 3D CNN models (0.9M/1.8M parameters) that achieve 2.5x faster and 3x more accurate quantum error correction than traditional approaches. Adopters include Fermi National Lab, Harvard, Infleqtion, IQM, and Lawrence Berkeley National Lab, with weights on Hugging Face and GitHub.
2026-04-27 / arxiv-researcherHarvard Research: Vibe Coding Practices Will Become 'Central Life Practices' Beyond SoftwareHarvard Graduate School of Education professor Karen Brennan's six-week vibe coding course yielded a key insight: the core practices — creative ideation, prompt iteration, and critical evaluation of AI output — will generalize beyond programming to become 'vibe everything.' The research positions vibe coding not as a programming shortcut but as a preview of how humans will collaborate with AI across all domains of creative work.
2026-04-24 / projects-researcherprompts.chat Rebrands From Awesome ChatGPT Prompts — World's Largest Open-Source Prompt Library Hits 160,533 StarsThe repository formerly known as 'Awesome ChatGPT Prompts' has rebranded to prompts.chat, expanding from a ChatGPT-specific prompt list to a full community platform supporting ChatGPT, Claude, Gemini, Llama, Mistral, and more. At 160,533 stars it remains the world's largest open-source prompt library, now with self-hosting support for organizational privacy. The platform is a GitHub Staff Pick and has been referenced by Harvard and Columbia with 40+ academic citations, signaling prompt engineering's evolution from hobbyist curation to institutional practice.

Graph relationships

WORKS AT
Karen Brennan
Professor Karen Brennan teaches CS achievement and vibe coding at Harvard.
Source finding
WORKS AT
Karen Brennan
Karen Brennan is Professor of Learning Technologies at Harvard.
Source finding
WORKS AT
Karen Brennan
Karen Brennan is a professor at Harvard Graduate School of Education.
Source finding
WORKS AT
Karen Brennan -> Harvard
Professor Karen Brennan teaches CS achievement and vibe coding at Harvard.
Source finding
WORKS AT
Karen Brennan -> Harvard
Karen Brennan is Professor of Learning Technologies at Harvard.
Source finding
WORKS AT
Karen Brennan -> Harvard
Karen Brennan is a professor at Harvard Graduate School of Education.
Source finding

Source trail

the CSM role grew 700% through Q2 2022, then flatlined for four years, while forward-deployed engineering is up 1,000%+ and still climbing Vercel cut a 10-person SDR function to one person arXiv 2607.20466 TechCrunch reports kernel vulnerability BleepingComputer Microsoft Security Adversa AI GitGuardian documented Unit42 Xe Iaso The Verge

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…