#AIhype take-downs

On reflecting on how much time I spend working to deflate and counteract AI hype, and how much of that ends up as Twitter threads, I thought it would be useful to have a place to link to all of them. Trying this out as a Medium post rather than just a page on my own website on the theory that this will have greater reach.

Links given in reverse-chronological order, to make it easier to find new things when readers check back. (Though in the short term, I may be turning up more older ones and slotting them in to their spot in the list.)

I’m only listing informal things here. For my publications see my publications page; for media appearances (including podcasts) see my media page. You may also be interested in Mystery AI Hype Theater 3000, my podcast with Alex Hanna (also available as videos).

• 4/2/24 Mystery AI Hype Theater newsletter post taking apart hype-by-proxy in the Chronicle of Higher Education
• 3/5/24 Parody of (practically self-parodying) SV angel open letter “Build AI for a Better Future” (original letter)
• 3/2/24 Thread on problematic reporting in The Guardian (Fediverse, Twitter, BlueSky) on the proliferation of robo-therapy apps.
• 2/29/24 Thread on uncritical FT reporting (Fediverse, Twitter, BlueSky) about UK gov’t plans to use generative AI to handle, inter alia, freedom of information requests.
• 2/15/24 Response (Fediverse, Twitter, BlueSky) to Kevin Roose’s one-year later piece on his reporting of his experience with “Sydney”
• 11/23/23 Response (Fediverse, Twitter, BlueSky) to Reuter’s reporting on “Q*”, AI safety and OpenAI.
• 5/27/23 Response (toot thread, tweet thread) to NPR’s coverage of Geoffrey Hinton’s AI doomerism tour.
• 5/3/23 Reponse (tweet thread) to Geoffrey Hinton saying that Timnit Gebru’s concerns are less “existentially serious” than his.
• 4/19/23 Dissection (toot thread, tweet thread) of Forbes informercial What The Technology Behind ChatGPT Can Do For Governance, Risk And Compliance
• 4/17/23 Response (tweet thread turned blog post) to Sundar Pichai claiming “emergent properties” for Bard on 60 minutes.
• 3/28/23 Response (toot thread, tweet thread) to open letter calling for a “pause” on “giant AI” experiments
• 2/25/23 Response (toot thread, tweet thread) to Sam Altman’s blog pot “Planning for AGI and Beyond”
• 1/16/23 Response (toot thread, tweet thread) to NYT op-ed “How ChatGPT Hijacks Democracy”
• 1/10/23 Response (toot thread, tweet thread) to NYT article “A.I. Turns Its Artistry to Creating New Human Proteins”.
• 12/29/22 Response (toot thread, tweet thread) to Freethink article claiming that chatbots “develop a near-precise understanding of the person on the other end”
• 12/15/22 Response (blog post) to hype-y coverage of ChatGPT in the Washington Post
• 12/14/22 Response (toot thread, tweet thread) to folks saying they are using ChatGPT to write academic papers.
• 11/16/22 Response (toot thread, tweet thread) to Meta AI’s announcement of Galactica & absurd claims that it is “an AI” and is useful for searching over scientific knowledge.
• 10/28/22 Analysis (tweet thread) of performance art (?) project trying to build a political party around a language model dubbed “Leader Lars”. (via threadreaderapp)
• 10/28/22 Response (tweet thread) to puff piece in the Globe and Mail about Cohere. (via threadreaderapp)
• 10/16/22 Analysis (tweet thread) of news coverage of research claiming that “AI spots signs of mental health issues in text messages on par with human psychiatrists” (via threadreaderapp)
• 10/2/22: Response (tweet thread) to Jack Clark’s claims that Anthropic’s language model “‘testified’ in the U.S. Senate”.
• 9/28/22: Response (tweet thread) to the OpenAI folks (and others) quoted in “AI can now create any image in seconds, bringing wonder and danger”, Nitasha Tiku, Washington Post.
• 9/27/22: Analysis (tweet thread) of Shotspotter in response to the news that Seattle is considering contracting with them.
• 9/23/22: Mystery AI Hype Theater (Video), with Alex Hanna, response to Blaise Agüera y Arcas’s Medium post “Can Machines Learn How to Behave?”, Part 3.
• 9/19/22: Response (tweet thread) to “Of God and Machines”, Stephen Marche, The Atlantic.
• 9/14/22: Response (tweet thread) to “10 years later, deep learning ‘revolution’ rages on, say AI pioneers Hinton, LeCun and Li”, Sharon Goldman, VentureBeat.
• 9/6/22: Mystery AI Hype Theater (Video), with Alex Hanna, response to Blaise Agüera y Arcas’s Medium post “Can Machines Learn How to Behave?”, Part 2.
• 8/31/22: Mystery AI Hype Theater (Video), with Alex Hanna, response to Blaise Agüera y Arcas’s Medium post “Can Machines Learn How to Behave?”, Part 1.
• 7/3/22: Response (tweet thread) to “Algorithm Claims to Predict Crime in US Cities Before It Happens” (original title “Algorithm Predicts Crime in US Cities Before It Happens”), Carrington York, Bloomberg.
• 5/25/22 Tweet thread reflecting on the cop-out that is “Making ethical NLP (or “AI”) systems is too hard because humans haven’t agreed on what is ethical/moral/right” and how that relates to fantasies of AGI.
• 4/28/22: Response (tweet thread) to Eric Jang’s blog post “All Roads Lead to Rome: The Machine Learning Job Market in 2022”
• 4/24/22: Response (tweet thread) to tweet thread from elicit.org claiming they had an “AI” that could do literature reviews
• 4/17/22: Response (blog post) to “A.I. Is Mastering Language. Should We Trust What It Says?”, Steven Johnson, New York Times Magazine.
• 4/7/22: Response (short tweet thread) to “Bot can spot depressed Twitter users in 9 out of 10 cases”, Tim Pilgrim, TechXplore.
• 4/6/22: Response (tweet thread) to “Can A.I.-Driven Voice Analysis Help Identify Mental Disorders?”, Ingrid K. Williams, New York Times.
• 3/16/22: Response (tweet thread) to Google blog post “Let’s get personal: adaptive learning tech and education”
• 1/20/22: Response (blog post) to Blaise Agüera y Arcas’s Medium post “Do large language models understand us?”
• 11/4/21: Response (tweet thread) to Microsoft trying to point to their own ethics guidelines in their introduction of GPT-3.
• 9/6/21: Tweet thread on large language models and why the problems are more than just chasing noise in the data.
• 8/28/21: Tweet thread reflecting on the effects of ML folks getting interested in NLP.
• 8/26/21: Tweet thread reflecting on how LLM research is sucking the oxygen out of the field of compling.
• 7/27/21: Tweet thread about Stanford HAI’s workshop on “universal models” (later renamed to “foundation models”).
• 5/21/21: Response (tweet thread) to “Prabhakar Raghavan Isn’t CEO of Google — He Just Runs the Place”, Steven Levy, WIRED.
• 1/8/21: Response (tweet thread) to “Microsoft’s AI model has outperformed humans in natural language understanding”, Usama Jawad, Neowin, and the Microsoft press release the piece is based on.
• 4/18/19: Response (tweet thread) to suggestions of using BERT or GPT-2 to “unredact” the Mueller report.