Hiya, folks, and welcome to TechCrunch\u2019s regular AI newsletter.<\/p>\n
This week in AI, generative AI<\/a> is beginning to spam up academic publishing \u2014 a discouraging new development on the disinformation front.<\/p>\n In a post on Retraction Watch<\/a>, a blog that tracks recent retractions of academic studies, assistant professors of philosophy Tomasz \u017buradzk and Leszek Wro\u0144ski wrote about three journals published by Addleton Academic Publishers\u00a0that appear to be made up entirely of AI-generated articles.<\/p>\n The journals contain papers that follow the same template, overstuffed with buzzwords like \u201cblockchain,\u201d \u201cmetaverse,\u201d \u201cinternet of things\u201d and \u201cdeep learning.\u201d They list the same editorial board \u2014 10 members of whom are deceased \u2014 and a nondescript address in Queens, New York, that appears to be a house.<\/p>\n So what\u2019s the big deal? you might ask. Isn\u2019t flipping through AI-generated spammy content simply the cost of doing business on the internet these days?<\/p>\n Well, yes. But the fake journals show how easy it is to game the systems used to evaluate researchers for promotions and hiring \u2014 and this could be a bellwether for knowledge workers in other industries. <\/p>\n On at least one widely used evaluation system, CiteScore, the journals rank in the top 10 for philosophy research. How is this possible? They extensively cross-cite each other. (CiteScore considers citations in its calculations.) \u017buradzk and Wro\u0144ski find that, of 541 citations in one of Addleton\u2019s journals, 208 come from the publisher\u2019s other fake publications.<\/p>\n \u201c[These rankings] frequently serve universities and funding bodies as indicators of the quality of research,\u201d \u017buradzk and Wro\u0144ski wrote. \u201cThey play a crucial role in decisions regarding academic awards, hiring and promotion, and thus may influence the publication strategies of researchers.\u201d<\/p>\n One could argue that CiteScore is the problem \u2014 clearly it\u2019s a flawed metric. And that\u2019s not a wrong argument to make. But it\u2019s also not wrong to say that generative AI and its abuse are disrupting systems on which people\u2019s livelihoods depend in unexpected \u2014 and potentially quite damaging \u2014 ways.<\/p>\n There\u2019s a future in which generative AI causes us to rethink and reengineer systems like CiteScore to be more equitable, holistic and inclusive. The grimmer alternative \u2014 and the one that\u2019s playing out now \u2014 is a future in which generative AI continues to run amok, wreaking havoc and ruining professional lives.<\/p>\n I sure hope we course-correct soon.<\/p>\n DeepMind\u2019s soundtrack generator<\/a><\/strong>: DeepMind, Google\u2019s AI research lab, says it\u2019s developing AI tech to generate soundtracks for videos. DeepMind\u2019s AI takes the description of a soundtrack (e.g., \u201cjellyfish pulsating under water, marine life, ocean\u201d) paired with a video to create music, sound effects and even dialogue that matches the characters and tone of the video.<\/p>\n A robot chauffeur<\/a><\/strong>: <\/strong>Researchers at the University of Tokyo developed and trained a \u201cmusculoskeletal humanoid\u201d\u00a0called Musashi to drive a small electric car through a test track. Equipped with two cameras standing in for human eyes, Musashi can \u201csee\u201d the road in front of it as well as the views reflected in the car\u2019s side mirrors.<\/p>\n A new AI search engine<\/a><\/strong>: <\/a><\/strong>Genspark, a new AI-powered search platform, taps generative AI to write custom summaries in response to search queries. It\u2019s raised $60 million so far from investors, including Lanchi Ventures; the company\u2019s last funding round valued it at $260 million post-money, a respectable figure as Genspark goes up against rivals like Perplexity.<\/p>\n How much does ChatGPT cost?<\/a><\/strong>: How much does ChatGPT, OpenAI\u2019s ever-expanding AI-powered chatbot platform, cost? It\u2019s a tougher question to answer than you might think. To keep track of the various ChatGPT subscription options available, we\u2019ve put together an updated guide to ChatGPT pricing.<\/p>\n Autonomous vehicles face an endless variety of edge cases, depending on the location and situation. If you\u2019re on a two-lane road and someone puts their left blinker on, does that mean they\u2019re going to change lanes? Or that you should pass them? The answer may depend on whether you\u2019re on I-5 or the Autobahn.<\/p>\n A group of researchers from Nvidia, USC, UW, and Stanford show in a paper just published at CVPR that a lot of ambiguous or unusual circumstances can be resolved by, if you can believe it, having an AI read the local drivers\u2019 handbook.<\/p>\n Their Large Language Driving Assistant, or LLaDa<\/a>, gives LLM access to \u2014 not even fine-tuning on \u2014 the driving manual for a state, country, or region.\u00a0Local rules, customs, or signage are found in the literature and, when an unexpected circumstance occurs like a honk, high beam, or herd of sheep, an appropriate action (pull over, stop turn, honk back) is generated.<\/p>\n It\u2019s by no means a full end-to-end driving system, but it shows an alternate path to a \u201cuniversal\u201d driving system that still encounters surprises. Plus perhaps a way for the rest of us to know why we\u2019re being honked at when visiting parts unknown.<\/p>\n On Monday,\u00a0Runway<\/a>, a\u00a0company building generative AI tools geared toward film and image content creators,\u00a0unveiled Gen-3 Alpha<\/a>. Trained on a vast number of images and videos from both public and in-house sources, Gen-3 can generate video clips from text descriptions and still images.<\/p>\n Runway says that Gen-3 Alpha delivers a \u201cmajor\u201d improvement in generation speed and fidelity over Runway\u2019s previous flagship video model,\u00a0Gen-2<\/a>, as well as fine-grained controls over the structure, style and motion of the videos that it creates. Gen-3 can also be tailored to allow for more \u201cstylistically controlled\u201d and consistent characters, Runway says, targeting \u201cspecific artistic and narrative requirements.\u201d<\/p>\n Gen-3 Alpha has its limitations \u2014 including the fact that its footage maxes out at 10 seconds. However, Runway co-founder Anastasis Germanidis promises that it\u2019s just the first of several video-generating models to come in a next-gen model family trained on Runway\u2019s upgraded infrastructure.<\/p>\n Gen-3 Alpha is only the latest generative video system of several to emerge on the scene in recent months. Others include OpenAI\u2019s Sora<\/a>, Luma\u2019s Dream Machine and Google\u2019s Veo<\/a>. Together, they threaten to upend the film and TV industry as we know it \u2014 assuming they can beat copyright challenges<\/a>.<\/p>\n AI won\u2019t be taking your next McDonald\u2019s order. <\/p>\n McDonald\u2019s this week announced<\/a> that it would remove automated order-taking tech, which the fast-food chain had been testing for the better part of three years, from more than 100 of its restaurant locations. The tech \u2014 co-developed with IBM and installed in restaurant drive-thrus \u2014 went viral last year for its propensity to misunderstand customers and make mistakes.<\/p>\n A recent piece<\/a> in the Takeout suggests that AI is losing its grip on fast-food operators broadly, who not long ago expressed enthusiasm for the tech and its potential to boost efficiency (and reduce labor costs). Presto, a major player in the space for AI-assisted drive-thru lanes, recently lost a major customer, Del Taco, and faces mounting losses.<\/p>\n The issue is inaccuracy. <\/p>\n McDonald\u2019s CEO Chris Kempczinski told<\/a> CNBC in June 2021 that its voice-recognition technology was accurate about 85% of the time, but that human staff had to assist with about one in five orders. The best version of Presto\u2019s system, meanwhile, only completes roughly 30% of orders without the help of a human being, according to the Takeout.<\/p>\nNews<\/h2>\n
Research paper of the week<\/h2>\n
![\"\"](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2024\/06\/nl-driveanywhere.jpg?w=680\")
Model of the week<\/h2>\n
Grab bag<\/h2>\n