OpenAI founder raises $1 billion for “Super Safe” AI

Ilya Sutskever, a revered AI scientist who was a founder of OpenAI, has now raised $1 billion for an independent venture aiming at making superintelligent AI systems, with abilities far exceeding humans, but are safe for human use. Sutskever was Chief Scientist at OpenAI, but left as part of the fallout of the chaotic Board coup that fired cofounder and CEO Sam Altman last November, only to be forced to resign themselves when Altman staged an employee revolt which brought him back as CEO within days. Sutskever sided first with the Board, and later with Altman, and ended up being odd man out. However battered Sutskever may have been by that experience, he now has 1 billion new green friends to console him, courtesy of star-struck VCs.

OpenAI founder raises $1 billion for “Super Safe” AI

Apple reveals AI features on new iPhones

On Monday Apple unveiled its new iPhone 16 line, which will serve as the platform to deliver Apple’s brand of AI. Apple, for a variety of reasons, including data security and privacy, has decided to run its own tiny AI model directly on the phone, and only offload complex tasks to larger models in the cloud on an as-needed basis. This means that, for now at least, Siri will only get a little bit smarter, no match for the flirty multi-tasking voice assistant demoed by OpenAI in May. Apple is hinting at bigger upgrades to Siri in the near future. Meanwhile, AI is being woven into many of the features of the new iPhone. The onboard AI can help you edit, rewrite, and summarize text in any app where you type. It will also help you find any photo on your phone from a text description. Photo editing is now turbocharged with AI, allowing you to easily remove extraneous objects (or people). And you can now point your camera at any storefront or street scene and the AI will help you identify what you are looking at. The competition in this space from Google and OpenAI is likely to be fierce, so expect rapid improvements (and a return to the days of needing to upgrade your phone annually to get the newest features.)

Apple reveals AI features on new iPhones

Google’s Waymo is nearly 4x safer than human drivers

Waymo, a driverless taxi service owned by Google parent Alphabet, operates in Phoenix and San Francisco. The company recently released a safety analysis of the first 20 million passenger miles driven by the company, and the numbers indicate that robot taxis are more than 3 times safer than human drivers.

Waymo autonomous taxis were involved in accidents causing injury at a rate of only 0.8 incidents per million miles driven. Human drivers, in contrast, were involved in 2.92 incidents per million miles. Waymo is committed to gradually expanding its service into additional metropolitan areas over the next few years.

Waymo’s robo-taxis are more than 3 times safer than human drivers

Salesforce deploys AI-in-a-box: just add water and stir

Customer relationship software giant Salesforce has made a big push into AI, and is now trying to package up what it has learned into easy-to-use AI apps for its customers to automate common tasks. Known as Industries AI, the AI bots can automate such tasks as matching patients with appropriate clinical trials, streamlining recruitment processes, and providing maintenance alerts for company vehicles and equipment. In all, Salesforce reports that its bots can automate over 100 common tasks for its customers. This mix-and-match, Lego-block approach to AI automation is designed to assist customers who do not have the resources or expertise to develop and implement AI solutions by themselves, and is likely to become an attractive model for deploying AI in small and medium businesses.

Salesforce deploys AI-in-a-box

Carolina man arrested for $10 million music streaming scam

Last week, the US Department of Justice made public indictments against a North Carolina musician, charging that he had used AI to create thousands of fake bands, thousands of fake songs, and thousands of bots to “listen” to the fake songs on music streaming services, all to fraudulently receive royalty payments that eventually totaled millions of dollars. By spreading the payments among so many songs and artists, the alleged perpetrator, Michael Smith, was able to stay under the radar of the streaming services’ fraud detection software, which looks for anomalous streaming behavior. Smith is facing multiple felony counts, each of which carries a maximum penalty of 20 years in prison. This level of epic criminality combined with complete obliviousness to the potential consequences may cause “Carolina man” to rival “Florida man” as a label for half-smart self-destructive dimwits.

Carolina man arrested for $10 million music streaming scam

Time’s 100 most influential people in AI

TIME magazine has released its annual “100 Most Influential People in AI”, and like last year, the list is filled with both the usual suspects, as well as a smattering of ringers that make you wonder what recreational pharmaceuticals the editors were smoking. Case in point, the most prominent figure on the cover is Jensen Huang, the CEO of premiere AI chipmaker Nvidia, which is currently one of the 3 most valuable companies in the world. So far, so good. Next to him, though, and almost as prominent visually, is actress Scarlett Johannson, whose only known connection to AI involves a kerfuffle in which she threatened to sue OpenAI over a bot voice that she alleged was a clone of her own. Also, inexplicably, Elon Musk is left out. No matter how delicious my schadenfreude is over this snub to the world’s richest man-baby, Elon was an original investor in OpenAI, has since created a highly capable (if quirky) AI model known as Grok, and as CEO of EV automaker Tesla is responsible for one of the largest AI automation projects in the world, the “Full Self Driving” semi-autonomous driving mode for the millions of Tesla cars. So yeah, the editors must have been into the goood stuff to forget about Elon.

YouTube develops tools to combat deepfakes and voice clones

The ability of AI to create realistic and convincing simulations of the faces and voices of real people is growing exponentially, posing a threat to public and private figures alike. YouTube is developing tools to identify content on its platform that is too similar to the appearance, voice, or music of its content creators. The tools are being put into a toolbox available to content creators, so that they can police content that infringes on their likeness, voice, or music. This approach to the problem is helpful, but it must be noted that YouTube is at least partially offloading its responsibility to maintain a safe environment on its platform to the victims of its failure to do so.

YouTube develops tools to combat deepfakes and voice clones

James Earl Jones’ Darth Vader voice immortalized with AI

Iconic actor James Earl Jones, known for both his thespian abilities and a unique rumbling bass voice, passed away on Monday at age 93. Prior to his final illness, Jones agreed to have his voice cloned by AI, so that future Star Wars dramas would be able to have an authentic simulacrum of his Darth Vader voice performance.

James Earl Jones’ Darth Vader voice immortalized with AI

Apple goes deeper into health tech

One of the important but less-noticed aspects of Apple’s unveiling of new products on Monday was the company’s doubling down on incorporating health technology into its products. The Apple Watch, already one of the best fitness trackers on the market, and FDA-approved to detect atrial fibrillation, has now been upgraded to detect sleep apnea. In addition, AirPods have been re-engineered to administer a hearing test, and double as a highly affordable alternative to conventional hearing aids. The move into health tech seems deliberate and sustained, so expect more products and features along these lines in the future.

Apple goes deeper into health tech

Google DeepMind announces Alpha Proteo drug designer

Google’s DeepMind AI research arm has just announced Alpha Proteo, an AI system that designs novel proteins that bind more successfully to target molecules. Protein binders have been found useful for binding to cancer targets, blocking viral infections, and modulating immune response. However, prior efforts to develop such protein binders involved a lot of hit-or-miss lab experimentation. DeepMind says that Alpha Proteo can shorten the time to discovery, while producing proteins that bind more tightly to the target molecule, which has favorable implications for efficacy. They are inviting researchers to collaborate on projects that would validate the utility of the Alpha Proteo system.

Google DeepMind announces Alpha Proteo drug designer