Google parent Alphabet will reorganize its AI unit as part of companywide layoffs announced today, Bloomberg reports. The company said it will eliminate 12,000 jobs across all units globally, representing 6.4% of its global workforce, representing the largest round of layoffs in its 25-year history. More: - In a memo, Google AI lead Jeff Dean said there has been a "modest" amount of layoffs in its research unit.
- The company also "over-invested" in Google for Clinicians, its health information data platform for health care teams, he noted. Reports did not specify how many in that unit were impacted.
- Citing sources, The Information notes that Google Brain, its AI lab, remains mostly unaffected. This unit is now working alongside product groups to advance more AI-powered services for the search giant, it said.
- Last week, Alphabet's Intrinsic robotics subsidiary initiated layoffs affecting nearly 20% of its workforce.
Zoom out: - In the memo, Dean noted that Google is still committed to its AI advancements. "Even in just the last year or so, together we've made great strides in our core research on advancing AI and other important areas," he wrote.
- On Friday, Alphabet said it will cut positions across the entire organization after it conducts reviews of its functions and product areas.
- This will help ensure that employees and roles are aligned with its highest priorities, said CEO Sundar Pichai, who today also emphasized artificial intelligence as an important area for the company moving forward.
BLOOMBERG 
| |
Facing growing competition from OpenAI, Google is ramping up its AI efforts with plans to announce over 20 new AI projects this year, according to The New York Times. Google is planning to show off a demo version of its search engine equipped with chatbot capabilities this year. Some of its products could be unveiled during Google's I/O developers conference in May. AI projects Google is working on: - An image generator that can both create and edit images.
- A code generation tool, PaLM-Coder 2, which could rival Microsoft's GitHub Copilot.
- An upgraded version of the AI Test Kitchen app, where users can interact with experimental AI technologies, including LaMDA, Google's language model for conversations.
- A "Colab + Android Studio" tool to help developers more easily create apps for Android phones.
- A tool that summarizes videos by creating new ones.
- A wallpaper maker for Pixel smartphones.
- A green screen mode for YouTube.
- A "Shopping Try-on" feature.
Related: - Google founders Larry Page and Sergey Brin have also been more heavily involved with Google lately, according to The Times.
- CEO Sundar Pichai reportedly requested to meet with the co-founders, who are controlling shareholders of Alphabet, to go over the search giant's AI plans following the release of ChatGPT.
- Google is said to be working to accelerate the approval of its coming AI products, a process that ensures that they are ethical and fair.
THE NEW YORK TIMES
| |
 A message from LINEARB Cut your dev team’s code-review time by 40% for free! gitStream is the free dev tool that automates away the pull request & code review problems plaguing cycle time in over 90% of engineering teams. How does gitStream allow developers to spend more time coding and less time on toil & rework? - It adds context to every PR, like estimated time to review
- Automatically assigns the right reviewer and number of reviewers to PRs
- Code review automation that auto-request changes
- Automates PR approvals & merges
Install for free
| |
During a StrictlyVC interview, OpenAI CEO Sam Altman commented on the future release of GPT-4, saying that "people are begging to be disappointed, and they will be." He declined to say when the company's next large language model will be released, commenting only that it will "come out at some point when we are confident we can do it safely and responsibly." More: - Altman said OpenAI doesn't have a model capable of actual artificial general intelligence — when machine intelligence can mimic most aspects of human intelligence — and "that's sort of what's expected of us."
- A rumor dating back to 2021 claimed that GPT-4 will have 100 trillion parameters, which would be a notable step up from GPT -3's 175 billion parameters.
- However, Altman squashed that claim, calling it "complete bull—."
- "The GPT-4 rumor mill is a ridiculous thing," he said. "I don't know where it all comes from."
Related: - While Altman may be attempting to downplay the public's expectations for the future model, he did provide an update on GPT-4 in 2021, though it's unclear how many of his remarks are still true.
- At the time, Altman said GPT-4 would be heavily focused on coding, which is "where the available compute is going."
- He claimed GPT-4 will be a text model and not multi-modal, when models combine data and knowledge from multiple modalities and sources, improving performance.
- GPT-4 would not be much larger than GPT-3, he said, but "will use way more compute."
- When GPT-5 comes out, it may be able to pass the Turing Test, according to Altman, who still added that this is "probably not worth the effort."
- He also said that GPT-4 will not have 100 trillion parameters and that such a model may never be needed.
- "They are getting much more performance out of smaller models," Altman noted at the time.
THE VERGE
| |
Tech-focused publication CNET is pausing its use of AI-written stories, according to The Verge. During a staff call today, CNET leadership told employees that the AI-generated articles will be on hiatus "for now." Background: - Leadership did not provide a reason for the pause, which comes after CNET faced criticism for its use of the AI system.
- CNET used the system to help write and compile around 75 stories, which were fact-checked and edited by humans.
- At least one of the AI articles contained multiple errors, and CNET was forced to update it with a lengthy correction.
- The CNET Money editorial team began quietly experimenting with the technology in November, though it appears employees and the general public were largely unaware of its use.
Update: - On Friday, CNET executive content VP Lindsey Turrentine reportedly told staff that the publication would be more transparent about AI moving forward.
- The AI content generator is a proprietary tool developed by Red Ventures, CNET's owner.
- The private equity firm's Bankrate and its subsidiary, CreditCards.com, will also put the AI stories on pause for now.
THE VERGE
| |
 A message from DELL Dell for Startups provides startup expertise from dedicated technology advisors and scalable solutions to ensure your business is always ready to grow. From MVP to IPO, our Startup Tech Advisors are specially trained for your unique needs. They understand startups and will get to know your business. From client systems to infrastructure hardware, we have your startup covered with the best in scalable technology. Whether you're looking for low latency performance, deep learning, machine learning, VDI, and more. We have a solution that’s just right for you. The dell EMC Poweredge r750 is a full-featured enterprise server that delivers outstanding performance for the most demanding workloads. Inside members can save up to 50% off on Dell orders when connected to a Startup Advisor. For more information, go to www.dell.com/inside or contact us at startups@dell.com. Update your infrastructure
| |
Weekly roundup — Other top AI stories you might have missed this week: - Microsoft initiated layoffs affecting 10,000 employees or about 4.5% of its global workforce. In a memo announcing the layoffs, CEO Satya Nadella noted that the world is experiencing the "next major wave of computing" through AI advancements, and Microsoft is "turning the world's most advanced models into a new computing platform."
- Microsoft also announced the broad availability of its Azure OpenAI Service, where cloud customers can access OpenAI tools such as GPT-3.5, Codex, DALL-E, and soon, ChatGPT. More Azure business customers can now incorporate the generative AI technologies into their own applications in the cloud.
- Three visual artists filed a class-action lawsuit against three companies that have released AI-powered art generators. The companies are Midjourney, Stable Diffusion creator Stability AI, and DeviantArt, which recently released the AI art generator DreamUp.
- Getty Images said it is suing Stable Diffusion creator Stability AI, alleging copyright violations. The stock photo giant accused Stability AI of processing millions of copyright-protected images in order to train Stable Diffusion, its text-to-image creator.
- A chatbot powered by artificial intelligence successfully passed a university exam. George Mason University professor Alex Tabarrok says the conversational bot, developed by the startup Anthropic, marginally passed the blind-graded exam covering law and economics.
| |
QUICK HITS - Giving dry a try this January? Do it for less. Experience great-tasting non-alcoholic beer without the hangover! Enjoy $10 off your first purchase.*
- Two large language models — OpenAI's ChatGPT and Google's Flan-PaLM — recently passed the U.S. Medical Licensing Examination (USMLE), according to two separate research papers.
- The University of Chicago's Polsky Center for Entrepreneurship and Innovation and Data Science Institute announced Transform, an accelerator program for companies focused on AI and data science.
- Italian startup Cap_able debuted a clothing collection that prevents AI facial recognition software from identifying the wearer. The knitted garments display so-called "adversarial patches," which are AI-created designs that purposely confuse the software.
- Customer trust is critical. Here's how to enhance your security without overextending your resources.*
*This is a sponsored post.
| |
Upcoming Events: - January 23 - Inside Coffee Break (Register Here)
- January 24 - Inside Coffee Break (Register Here)
- January 25 - Amanda Natividad speaks on zero-click content, how to measure it, and other trends to keep tabs on in 2023. (Register Here) *
- January 26 - Growth Experimentation Workshop: From Strategy to Process by GrowthHackers.com (Register Here)
- January 29 - Inside Interview with Lacework - 4 Common Attack Paths in Cloud Security (Register Here)
- January 31 - AMA with LinearB - Improving workflow for developers w/ Ori Keren (Watch On Demand)
- January 31 - AMA with Gun.io - Building and managing software development teams w/ Deividi Silva (Watch On Demand)
- April 18 - Human Resources Summit'23 (Register Here)
*This is a sponsored listing.
| |
 |  | Beth is a writer and editor at Inside.com covering artificial intelligence and daily technology news. You can reach her at Beth.Duckett@inside.com. |   | | Editor | Aaron Crutchfield is based in the high desert of California. Over the last two decades, he has spent time writing and editing at various local newspapers and defense contractors in California. When he's not working, he can often be found looking at the latest memes with his kids or working on his 1962 and 1972 Fords. | |
| |
| |
Comments
Post a Comment