| |
| Here's a look at today's AI briefing: - Humana accused of using flawed AI to deny patient care.
- Google's Gemini Pro AI model rolls out for developers and enterprises.
- The White House AI Council holds first meeting.
- Research suggests ChatGPT hasn't raised school cheating rates.
Beth
p/beth-duckett | |
| 1 | | Health insurance company Humana is accused of using an AI model to deny rehab care to elderly patients. A lawsuit filed this week in the U.S. District Court for the Western District of Kentucky claims the AI was used to deny care to elderly Medicare Advantage patients, overriding some doctors' recommendations. More: - The lawsuit, which is seeking class-action status, claims Humana knew that its AI model was "highly inaccurate."
- Even so, the insurer has still used the AI to reject claims, knowing that only a small percentage of policyholders would appeal, the complaint alleges.
- The suit labels the AI as a "fraudulent scheme," leaving elderly patients with medical debt and/or without care.
Zoom out: - It's the second lawsuit to target nH Predict, an AI algorithm developed by NaviHealth that forecasts how long patients should stay in rehab.
- In November, families of two deceased patients sued UnitedHealth Group over the same AI model, saying it was used to "systematically deny" claims for Medicare patients recovering from debilitating illnesses in nursing homes.
- The complaint argues that nH Predict has a 90% error rate.
|     | |
| 2 | | Google is now offering its most advanced text-generating AI model, Gemini, to developers and businesses to create their own chatbots and apps. Gemini Pro, the mid-sized model in the Gemini lineup, is now available in two Google Cloud products: Google AI Studio and Vertex AI. More: - Gemini is available in three sizes: Ultra, Pro, and Nano. The first version of Gemini Pro has a 32K context window and is limited to text input/output.
- Google also announced a new Gemini Pro Vision multimodal model that accepts both text and images, though it generates only text.
- Pro is launching in Google AI Studio, its free web-based tool for individual developers and teams, as well as in Vertex AI, its cloud programming tool for enterprises to fine-tune AI models.
- Gemini Pro and Gemini Pro Vision are currently free for developers in AI Studio and on Vertex AI, with a rate limit of up to 60 requests per minute.
- Once the free period is over, Gemini Pro will be cheaper to use than GPT-4.
More announcements: - Google is also bringing its image generator, Imagen 2, to more Vertex AI customers. Imagen 2 has improved photorealism, text rendering, and logo generation over its predecessor.
- Google also announced the general availability of Duet AI for Developers, a coding tool with generative AI. Gemini will be integrated into Duet.
- The company introduced MedLM, a family of models tailored for the healthcare industry.
| | | |
| |
|
| 3 | | The White House AI Council held its first meeting on Tuesday, bringing together top government officials who discussed ways to carry out President Biden's AI executive order signed in October. The group includes cabinet members like Secretary of Commerce Gina Raimondo, Secretary of State Antony Blinken, Health and Human Services Secretary Xavier Becerra, and Director of National Intelligence Avril Haines. More: - The meeting involved a classified intelligence briefing on AI's capabilities from Biden's national security team, a source told The Hill.
- The council also addressed talent recruitment, safe testing of new models, and mitigating AI risks, including fraud, discrimination, and privacy concerns.
- Members talked about the new U.S. AI Safety Institute designed to help gauge the risks of advanced AI models and guide future rules.
Zoom out: - The group, created through Biden's AI executive order, will meet regularly.
- Under the executive order, any company creating an AI model with possible national security risks must report it to the government and give information on the testing and security measures taken.
- The order will also require federal agencies to assess large AI models before using them and share reports on how they plan to incorporate AI.
| | | |
| 4 | | Despite concerns about academic cheating, new Stanford University research suggests that AI chatbots like ChatGPT haven't significantly raised high school cheating rates. Pew Research Center also conducted a survey of U.S. teenagers and found that only 19% of teens who know about ChatGPT report using it for schoolwork. 
Details: According to Stanford, surveys of over 40 U.S. high schools this year found that cheating rates have remained about the same as in prior years, suggesting that ChatGPT isn't leading to widespread cheating as previously imagined. What the numbers show: In total, around two-thirds of U.S. teens are familiar with ChatGPT, with only 23% having heard "a lot" about it. Pew Research Center found that only 13% of all U.S. teens have used ChatGPT in their schoolwork. Teens in higher grade levels, specifically 11th and 12th graders, were the most likely to have used the AI chabot. When asked what are acceptable ways for students to use ChatGPT, nearly 70% of teens who use it said it's acceptable for researching new topics. The figures were lower for using it to solve math problems or write essays. The bigger picture: ChatGPT's debut last year triggered debates about its role in schools, with questions on whether it should be welcomed or banned in classrooms. Jeffrey Gottfried, an associate research director at Pew, noted that while most teens have some awareness about ChatGPT, it "is not a majority of teens who are incorporating it into their schoolwork quite yet." | | | |
| A message from our sponsor, 10 East. |  | | Where are sophisticated investors accessing high-quality deal flow for their personal portfolios? Enter 10 East. It’s a platform where qualified individuals can invest in vetted private credit, real estate, niche venture/private equity, and other one-off investments typically unavailable through traditional channels. Led by Michael Leffell, former Deputy Executive Managing Member of Davidson Kempner, the principals have a decade+ track record of strong performance across +355 transactions under their belts. And the best part? You have the flexibility to participate at your discretion, on a deal-by-deal basis, while co-investing alongside industry veterans who have skin-in-the-game. 10 East is where founders, partners, and portfolio managers from leading investment firms are diversifying their personal wealth. Join them with complimentary access here. Try it Now | |
|
| 5 | | Harvey, which specializes in generative AI for law firms, is in discussions to raise $70M to $80M in a funding round that values the AI startup at $700M, sources told The Information. The legal tech startup last raised $21M in a Series A round, led by Sequoia Capital, in April at a $150M valuation. More: - Kleiner Perkins, along with solo investor Elad Gil, will co-lead the latest funding round, with participation from existing investors Sequoia Capital and OpenAI Startup Fund.
- Harvey builds custom LLMs for law firms and developed a natural language platform that answers legal questions and performs legal research.
- Attorneys can use Harvey's interface to describe the task they want — such as rewriting a clause — and Harvey will generate an output.
Zoom out: - London law firm Macfarlanes and global firm Allen & Overy have both signed up to use Harvey's technology.
- In 2022, OpenAI's startup fund led a $5M round in Harvey.
| | | |
| 6 | | 🍋MORE FRESH FUNDING 🍋 - Andreessen Horowitz and Lightspeed Ventures led a €385M ($415M) funding round into French AI startup Mistral AI, valuing the seven-month-old startup at $2B. Salesforce, BNP Paribas, CMA CGM, General Catalyst, Elad Gil, Emerson Collective, Conviction, Bpifrance, and others participated in the round.
- Armada, an edge computing and data network startup launched by former DataRobot CEO Dan Wright, emerged from stealth with a $55M funding that valued the firm at about $250M. Founders Fund, Lux Capital, Shield Capital, and 8090 Industries led the round.
- Indian generative AI startup Sarvam raised $41M in Series A funding led by Lightspeed Venture Partners. Institutional investors participating in the round include Khosla Ventures and Peak XV Partners.
- AI-powered robotically-assisted imaging solutions developer Laza Medical secured $36M capital from investors. The Series A round was led by The Capital Partnership, with additional support from GE Healthcare, PA MedTech VC fund, Unorthodox Ventures, and Dara Holdings.
- Australian generative AI startup Leonardo.AI bagged $31M capital from Blackbird, Side Stage Ventures, Smash Capital, TIRTA Ventures, Gaorong Capital, and Samsung Next.
For the latest on AI startup funding and more in the VC community, subscribe to our Inside Venture Capital newsletter. | | | |
| 7 | | QUICK HITS - Want to master Kubernetes? The Kubernetes Cookbook is a must-have guide to discover guidance on cloud architecture analytics, optimization, governance, and more.*
- The U.S. is leading a new diplomatic effort at the UN to garner global support for "AI for good and for all."
- GM hired Anantha Kancherla, a former engineer at Meta's Reality Labs, as its VP of advanced driver-assistance systems. The automaker also introduced a website to showcase its AI endeavors.
- An artwork contest for the Wichita, Kansas Riverfest has sparked controversy for choosing a winning poster that was created with help from AI.
- Bestselling book on alternative investing offers the first chapter free for a limited time. "Money Shackles" innovates the way you think about money.*
*This is a sponsored post. | | | |
 | AI and technology writer | | Beth is a contributing editor and writer of Inside's AI and Tech newsletters. She has written for notable publications like USA Today and the Arizona Business Gazette. During her time as a public policy reporter at The Arizona Republic, she received recognition with a Pulitzer Prize nomination and a First Amendment Award for her collaborative reporting on pension cost increases. Beth also authored a book on the solar photovoltaic industry in 2016. You can reach her at Beth.Duckett@yahoo.com. | | This newsletter was edited by Aaron Crutchfield | |
| |
 | Developers bring their ideas to life with Docker. Accelerate how you build modern apps. | |
|
| |
|
| |
Comments
Post a Comment