“Latest Updates: Google Debuts Fresh Open LLMs, Rivian Enacts Employee Cutbacks, and Signal Introduces Usernames”

Gettyimages 1230444599 1 E1660816546129
Google releases new open LLMs, Rivian lays off staff and Signal rolls out usernamesWelcome, folks, to Week in Review (WiR), TechCrunch’s regular newsletter covering noteworthy happenings in the tech industry. This week, Google launched two new open large language models, Gemma 2B and Gemma 7B, in its continued bid for generative AI dominance. The company, which describes the LLMs as “inspired by Gemini,” its flagship family of GenAI models, made each available for commercial and research usage. Change Healthcare hit: Change Healthcare, one of the largest healthcare tech companies in the U.S., confirmed that a cyberattack on its systems occurred recently. YouTube triumphant: YouTube dominates TV streaming in the U.S., per Nielsen’s latest report.

Exploring AI: Combatting Racial Bias in Image Generating Technology

Gettyimages 1062086882
This week in AI, Google paused its AI chatbot Gemini’s ability to generate images of people after a segment of users complained about historical inaccuracies. Google’s ginger treatment of race-based prompts in Gemini didn’t avoid the issue, per se — but disingenuously attempted to conceal the worst of the model’s biases. Yes, the data sets used to train image generators generally contain more white people than Black people, and yes, the images of Black people in those data sets reinforce negative stereotypes. That’s why image generators sexualize certain women of color, depict white men in positions of authority and generally favor wealthy Western perspectives. Whether they tackle — or choose not to tackle — models’ biases, they’ll be criticized.

Court Filing Reveals Struggle of DuckDuckGo and Neeva in Competition Against Google’s Dominance in Search Startups

Gettyimages 828901244
Google’s proposed “Findings of Fact” filing documents the history of search competition, including Google’s own beginnings, its innovations, the competitive landscape, Google’s search ads business, distribution agreements, and more. When shareholders sold stock to various VC firms, those funds were not used to improve the search engine, the filing argues. But it contradicts this point, too, noting that a third of DuckDuckGo’s 50 employees in 2018 were working on improving the search engine, for example. The search engine receives only about 2.5% of general search queries in the U.S., despite estimates that 10% of people in the U.S. claim to be users. This, DuckDuckGo’s leadership had explained, is due to the fact that people often use its search engine for some, but not all of their search queries.

“Google’s Image-Generating AI: A Confession of Loss of Control”

Adobe Firefly Dogwalkers Tilt Blur
Google has apologized (or come very close to apologizing) for another embarrassing AI blunder this week, an image generating model that injected diversity into pictures with a farcical disregard for historical context. While the underlying issue is perfectly understandable, Google blames the model for “becoming” over-sensitive. But if you ask for 10, and they’re all white guys walking goldens in suburban parks? Where Google’s model went wrong was that it failed to have implicit instructions for situations where historical context was important. These two things led the model to overcompensate in some cases, and be over-conservative in others, leading to images that were embarrassing and wrong.

Google to phase out Google Pay app in US by year’s end

Gettyimages 1207206237
Google is sunsetting the Google Pay app in the US later this yearGoogle has announced that Google Pay is shutting down in the United States in June, as the standalone app has largely been replaced by Google Wallet. Users can continue to access the app’s most popular features right from Google Wallet, which Google says is used five times more than the Google Pay app in the United States. After June 4, users will no longer be able to send, request or receive money through the U.S. version of the Google Pay app. Users who used the Google Pay app to find offers and deals can still so do using the new deals destination on Google Search, the company says. Google says millions of people in over 180 countries use Google Pay to check out when shopping on desktop, mobile and in store.

2024: End of Gmail’s Simple HTML View – Google’s Plan to Revamp the Interface

Gmail Icon 2020 Ios
Most will have been defaulted to the “new” Gmail view long ago, so unless you have been specifically requesting the “basic HTML” view, nothing should change for you. The company is sunsetting Gmail’s basic HTML view, which allows users to look at their emails in a bare-bones state, starting January 2024. “We’re writing to let you know that the Gmail Basic HTML view for desktop web and mobile web will be disabled starting early January 2024. The Gmail Basic HTML views are previous versions of Gmail that were replaced by their modern successors 10+ years ago and do not include full Gmail feature functionality,” the email reads. The HTML version lacks a lot of features such as chat, spell checker, search filters, keyboard shortcuts, and rich formatting.

Google Pay Expands QR Payment Offerings to Local Indian Merchants Following Successful Trial Period

Gettyimages 1169635369
Google said Thursday it plans to roll out the SoundPod, its portable speaker designed to instantly validate and announce successful payments, to small merchants across India over the coming months. The Google Pay expansion in India, where the company is among the mobile payment market leaders, comes even as the firm winds down some of its payments apps in the U.S. The sound-box was invented to serve small Indian merchants unable to afford regular point-of-sale devices but accepting of UPI payments. The company said merchants who use SoundPod to process 400 payments in a month will get $1.5 in cashback. Reliance, India’s largest firm by market cap, also began testing a similar device at its campus last year, TechCrunch earlier reported.

Gemini-Powered AI Writing Tool Now Built-In to Chrome

Gettyimages 981522310
Google Chrome is getting a new AI writing generator today. At its core, this Gemin-powered tool is essentially the existing “Help me write” feature from Gmail, but extended to the entire web and powered by one of Google’s latest Gemini AI models. To get started, head to the Chrome settings menu and look for the ‘Experimental AI’ page. From there, you can easily enable the new writing feature, as well as Google’s new automatic tab organizer (which I haven’t found particularly useful or smart so far) and the new Chrome theme manager). If you’re subscribed to Gemini Advanced, this new tool will not give you access to an enhanced writing model, a Google spokesperson told us.

Google halts AI system Gemini’s image generation of humans following historical errors

Glitched Google Logo
“While we do this, we’re going to pause the image generation of people and will re-release an improved version soon,” it added. While we do this, we're going to pause the image generation of people and will re-release an improved version soon. — Google Communications (@Google_Comms) February 22, 2024Google launched the Gemini image generation tool earlier this month. Gemini’s Al image generation does generate a wide range of people. An earlier AI image classification tool made by Google caused outrage, back in 2015, when it misclassified black men as gorillas.

Introducing the Formation of Google DeepMind’s AI Safety Organization

So in response, Google — thousands of jobs lighter than it was last fiscal quarter — is funneling investments toward AI safety. This morning, Google DeepMind, the AI R&D division behind Gemini and many of Google’s more recent GenAI projects, announced the formation of a new organization, AI Safety and Alignment — made up of existing teams working on AI safety but also broadened to encompass new, specialized cohorts of GenAI researchers and engineers. But it did reveal that AI Safety and Alignment will include a new team focused on safety around artificial general intelligence (AGI), or hypothetical systems that can perform any task a human can. The AI Safety and Alignment organization’s other teams are responsible for developing and incorporating concrete safeguards into Google’s Gemini models, current and in-development. One might assume issues as grave as AGI safety — and the longer-term risks the AI Safety and Alignment organization intends to study, including preventing AI in “aiding terrorism” and “destabilizing society” — require a director’s full-time attention.