Cruise's valuation halved after its driverless car hit and dragged a woman

Plus: Google CEO admits company got it wrong when it built 'too-woke' AI, and more

AI in brief Cruise, the self-driving biz backed by General Motors, had its valuation slashed by more than half since one of its cars crashed and dragged a woman down a street.

The woman was initially knocked over by a hit and run driver in another lane, throwing her into the path of a Cruise autonomous vehicle, which also hit her and then dragged her beneath its wheels.

The internal per-share price for the company estimated by a third party is said to be at $11.80, compared to $24.27 last quarter, company employees were reportedly told in an email. "We cannot ignore that this estimate is significantly lower than we've seen before and that there are real life impacts for each of us," Craig Glidden, chief administrative officer for Cruise, said in the email seen by Reuters. 

In 2021, after Cruise raised $8 billion in investments from Microsoft, Walmart, and top investors, it was valued at a whopping $30 billion. But over time, the hype for autonomous cars has died a little and Cruise's October disaster decreased its valuation. 

The company has now grounded all of its fleets across the US as it deals with the aftermath of the incident, which left the woman severely injured.

Officials from the California Public Utilities Commission, the National Highway Traffic Safety Administration, the Department of Justice, and the Securities Exchange Commission have launched investigations into the accident. 

'We got it wrong,' says Google CEO

Google's latest Gemini controversy wiped $90 million from its market value in the past week, and CEO Sundar Pichai told staff the mishap is "completely unacceptable." 

The company has become a laughing stock over its AI model being overly cautious with generating images of White people. In a bid to make Gemini less biased, unlike most systems tend to be, it overcompensated to make more diverse images depicting more women and people of color. 

Users, however, were frustrated and amused when Gemini produced unrealistic images of popes, Vikings, ice hockey players, and so on. The Founding Fathers of US, for example, were portrayed as being Black and Asian men even though they were of European descent.

"I know that some of its responses have offended our users and shown bias – to be clear, that's completely unacceptable and we got it wrong," Pichai said in a company meeting. Engineers are "working around the clock" to fix the issue.

"Our mission to organize the world's information and make it universally accessible and useful is sacrosanct. We've always sought to give users helpful, accurate, and unbiased information in our products. That's why people trust them. This has to be our approach for all our products, including our emerging AI products," Pichai added.

That's all the more reason for its models to be accurate. 

Tumblr and Wordpress in talks with OpenAI for selling user content

Blogging sites Tumblr and Wordpress are reportedly in talks to license users' published content to OpenAI and Midjourney.

Both companies develop commercial tools capable of generating synthetic content given input text-based descriptions, and require more and more data to train their AI models. Publishing platforms are increasingly striking up deals with LLM developers to receive monetary compensation for their data.

The source of content on Tumblr and Wordpress, however, is particularly controversial, considering users may not want their thoughts, photographs, and more to be ingested by AI. OpenAI, Midjourney, Google, Meta and more often believe that any public data is fair game to be scraped. But recent copyright infringement lawsuits have forced companies to be careful and many are now negotiating contracts to license data. 

The compilation process of huge datasets is messy. Cyle Gage, a product manager at Tumblr, said the company had collected public data uploaded to its platform from 2014 to 2023, but couldn't completely scrub private posts, NSFW content, or text belonging to third-parties, 404 Media reported. It's not clear whether OpenAI and Midjourney have gotten their hands on this or not, though.

Hardware and telco giants team up to develop AI for the edge

Top hardware makers, academics, and telecommunications companies are banding together to form AI-RAN Alliance, a new industry group focused on developing AI edge applications on 6G. 

Members of AI-RAN Alliance include: AWS, Arm, DeepSig, Ericsson, Microsoft, Nokia, Northeastern University, Nvidia, Samsung, SoftBank, and T-Mobile so far. Most AI models are too large to run locally on devices, making technologies like generative AI limited on smartphones and more. 

The group wants to advance radio access networks (RAN) to support running AI workloads on 5G and 6G-powered devices. More specifically, it is aiming to improve spectral efficiency, integrate AI and RAN more effectively in the hopes of 'generating new AI-driven revenue opportunities'.

"AI will fundamentally change the way wireless services are deployed and enable broad innovation and operational efficiency across the telco sector," Arm's Mohamed Awad, Senior Vice President and General Manager, Infrastructure Line of Business, Arm, said in a statement this week. "The AI-RAN Alliance brings together industry-shaping companies with expertise from silicon through software to deliver on the promise of ubiquitous AI and 6G." ®

More about

TIP US OFF

Send us news


Other stories you might like