Toggle light / dark theme

CB Insights has unveiled the winners of the seventh annual AI 100 — a list of the 100 most promising private AI companies across the globe.

Around one-third of this year’s winners are focused on AI applications across specific industries — such as visual dubbing for the media & entertainment sector or textile recycling for fashion & retail. A total of 40 vendors are focused on cross-industry solutions, like AI assistants & human-machine interfaces (HMIs), digital twins, climate tech, and smell tech.

Additionally, 27 companies in this cohort are developing tools like vector database tech and synthetic datasets to support AI development.

“There are many exciting use cases for generative speech models, but because of the potential risks of misuse, we are not making the Voicebox model or code publicly available at this time,” the company said in a research post. “While we believe it is important to be open with the AI community and to share our research to advance the state of the art in AI, it’s also necessary to strike the right balance between openness with responsibility.”

Representatives for Meta did not immediately respond to Insider’s request for comment, made outside normal working hours.

Meta said in the news release that the model could allow visually impaired people to hear messages from friends in their voices or allow users to speak in foreign languages in their own voice. The company also said the tech opened up the possibility for creators to edit audio tracks for video or create more natural-sounding voices for virtual assistants.

This week, a team of over 1,000 scientists from around the globe released to the public the first batch of data collected with the Dark Energy Spectroscopy Instrument (DESI), a telescope that cosmologists hope will help answer open questions on the nature of dark energy and the evolution of the Universe [13]. “The telescope works better than we ever imagined,” says Michael Levi, a cosmologist at Lawrence Berkeley National Laboratory (LBNL), California, and the director of the DESI Collaboration. “We are ready to have everybody look at this [initial] data release and see what they can do with it.”

The goal of the five-year-long DESI survey is to map the Universe deeper in time and higher in detail than any previous telescope (see Feature: Entering a New Era of Dark Energy Cosmology). “We want to go way beyond what was done before and really be able to see the evolution of dark energy over the history of the Universe,” says Nathalie Palanque-Delabrouille, a cosmologist at LBNL and one of the spokespeople for the DESI Collaboration. To see that evolution, the survey plans to pinpoint the locations of over 40 million galaxies. The key to filling in the cosmic map is the use of robotic technology that automatically alters the placements of light-collecting fibers so that they can retrieve spectroscopic information from targeted bright spots in the sky. The spectral measurements provide information on what an object is and how fast it is moving away from us, which is needed to estimate its distance.

The robotic technology used to target objects had never been tried before, so it was not always clear that DESI would perform as expected, Levi says. But he and other team members have been pleasantly surprised by how smoothly the machine has operated. “DESI has preserved every photon that the Universe gave us,” he says.

Join journalist Pedro Pinto and Yuval Noah Harari as they delve into the future of artificial intelligence (A.I.). Together, they explore pressing questions in front of a live audience, such as: What will be the impact of A.I. on democracy and politics? How can we maintain human connection in the age of A.I.? What skills will be crucial for the future? And what does the future of education hold?

Filmed on May 19 2023 in Lisbon, Portugal and produced by the Fundação Francisco Manuel dos Santos (FFMS), in what marks the first live recording of the show: “It’s not that simple.”

Don’t forget to subscribe to Yuval’s Channel, where you can find more captivating content!
@YuvalNoahHarari.

Stay connected with Yuval Noah Harari through his social media platforms and website:

This week my guest Anne Scherer, a professor of marketing at the University of Zurich who specializes in the psychological and societal impacts that result from the increased automation and digitization of the consumer-company relationship.

In this episode we focus on the details Anne covers in, You and AI, a book she co-authored with Cindry Candrian to bring an accessible understanding of the ways in which AI is shaping our lives. This takes on a tour of topics such as our symbiotic relationship with AI, manipulation, regulation, the proposed 6 month pause on AI development, the business advantages of better data policies around AI, the difference between artificial intelligence and human intelligence, and more.

Find out more about Anne and her book at annescherer.me.

Host:⁠⁠⁠⁠⁠⁠⁠ Steven Parton⁠⁠⁠⁠⁠⁠⁠ — ⁠⁠⁠⁠⁠⁠⁠LinkedIn⁠⁠⁠⁠⁠⁠⁠ /⁠⁠⁠⁠⁠⁠⁠ Twitter⁠⁠⁠⁠⁠⁠⁠

Music by: Amine el Filali.

————–

The field of photonics has seen significant advances during the past decades, to the point where it is now an integral part of high-speed, international communications. For general processing photonics is currently less common, but is the subject of significant research. Unlike most photonic circuits which are formed using patterns etched into semiconductor mask using lithography, purely light-based circuits are a tantalizing possibility. This is the focus of a recent paper (press release, ResearchGate) in Nature Photonics by [Tianwei Wu] and colleagues at the University of Pennsylvania.

What is somewhat puzzling is that despite the lofty claims of this being ‘the first time’ that such an FPGA-like device has been created for photonics, this is far from the case, as evidenced by e.g. a 2017 paper by [Kaichen Dong] and colleagues (full article PDF) in Advanced Materials. Here the researchers used a slab of vanadium dioxide (VO2) with a laser to heat sections to above 68 °C where the material transitions from an insulating to a metallic phase and remains that way until the temperature is lowered again. The μm-sized features that can be created in this manner allow for a wide range of photonic devices to be created.

What does appear to be different with the photonic system presented by [Wu] et al. is that it uses a more traditional 2D approach, with a slab of InGaAsP on which the laser pattern is projected. Whether it is more versatile than other approaches remains to be seen, with the use of fully photonic processors in our computers still a long while off, never mind photonics-accelerated machine learning applications.