| Here's a look at today's AI briefing: - Microsoft engineer contacts FTC with safety concerns about AI image creator.
- Study: U.S. trust in AI companies falls to 35%.
- Meta AI model improves its video recommendations.
- Microsoft to showcase AI for businesses at March 21 event.
Beth
p/beth-duckett | |
| 1 | | A Microsoft AI engineer has raised concerns with the FTC about the company's image-generator tool, saying it can create unsafe, violent, and sexual imagery. Shane Jones, a principal software engineering manager, has spent months bringing up safety concerns about Microsoft's Copilot Designer, which is built on OpenAI's DALL-E 3. More: - Jones, a six-year Microsoft veteran, says he found flaws and security risks in the AI image generator since December, allowing it to produce offensive images.
- Jones shared his concerns with U.S. Senate staff last month and sent letters to the Federal Trade Commission and Microsoft's board this week.
- In the FTC letter, Jones writes that Copilot Designer can create images featuring underage drinking and drug use, political bias, sexualized and violent images of women, conspiracy theories, religion, corporate trademarks, and more.
- He says he has urged Microsoft to withdraw Copilot Designer from the public until stronger safeguards are in place.
Zoom out: - In response, Microsoft said it appreciates Jones' efforts to test and study its technology to improve safety and is committed to tackling employee concerns about its policies.
- A spokesperson told The Verge that Microsoft has feedback and internal reporting tools to test and fix issues and has held "meetings with product leadership and our Office of Responsible AI to review these reports."
|     | |
| 2 | U.S. trust in AI companies has fallen from 50% to 35% over the last five years, according to a new Edelman study. The public relations firm found that trust in AI companies is low across the board, with Democrats at 38%, independents at 25%, and Republicans at 24%. 
What the numbers show: Global trust in AI companies has fallen from 61% to 53% over the past five years. Trust in the tech industry stands at 76%, while AI industry trust lags at 50%. What it means: Key reasons for the public's falling trust in AI companies are privacy concerns, fears of unchecked advances, a perceived lack of transparency, and concerns it could diminish human value, said Edelman's Justin Westcott. The bigger picture: While governments in the U.S. and the world are working on guidelines and laws to regulate AI, the public is generally dissatisfied with these efforts and is calling for "regulators to meet the public's expectations head-on," Westcott said. | | | |
| 3 | | Meta is working on a single advanced AI model to improve its video recommendations across its platforms, according to Facebook head Tom Alison. The new AI-driven engine is part of Meta's "technology roadmap" planned through 2026, he said. More: - Meta is already using the new engine for its TikTok-like Reels and plans to expand it to other video sections.
- According to Alison, testing the new AI-driven recommendation engine with Reels increased watch time by about 8% to 10%.
- He noted the model's architecture learns from data more efficiently than the prior version.
- Based on that success, Meta is working on the project to power video recommendations across Reels, Facebook's feed, and Groups using that single AI model rather than separate systems.
Zoom out: - In phase 1 of its technology roadmap, Meta transitioned its recommendation systems from conventional chips to GPUs.
- It's now in the third phase, which involves testing and using more GenAI and LLMs to power its products.
| | | |
| 4 | | Microsoft will showcase upcoming generative AI devices and features during an event on March 21. The "Advancing the New Era of Work with Copilot" event will focus on Microsoft's Surface, Windows, and Copilot products for businesses. The digital event will start streaming at 9 a.m. PT/noon ET on March 21. More: - The event page tells customers to tune in "for the latest in scaling AI in your environment with Copilot, Windows, and Surface."
- Microsoft is expected to unveil its new Surface Pro 10 tablet and Surface Laptop 6, which the company had dubbed its first "AI PCs."
- The commercial models have neural processing units (NPUs) to power AI features, including real-time captions and game upscaling, that are coming to Windows 11.
- Microsoft is also planning a new "AI Explorer" for Windows 11 for users to search across their PCs using natural language prompts.
| | | |
| |
|
| 5 | | Salesforce is launching an AI assistant that can automate some administrative tasks for healthcare workers. Doctors can use "Einstein Copilot: Health Actions" to help book appointments, summarize patient data, and send referrals using AI. More: - Salesforce says the AI software allows doctors to quickly generate patient summaries, including medications, diagnoses, and tests, to save time on manual data lookup.
- The tech giant is also launching "Assessment Generation" for medical offices to digitize health assessments without having to type or code them manually.
- The offerings are part of Salesforce's Einstein 1 Platform, which unifies medical data from sources like electronic health records and insurance companies into a single place.
| | | |
| 6 | | 🍋 FRESH FUNDING 🍋 - RapidSOS, a startup building AI-powered platform for emergency response teams, closed its Series C funding round at $150M, with an additional $75M funding from BlackRock.
- AI-powered image generation tool Ideogram secured a Series A funding of $80M. Andreessen Horowitz led the round, joined by Index Ventures, Redpoint Ventures, Pear VC, and SV Angel.
- Overjet, a startup developing an AI platform that helps diagnose disease from scans, secured $53.2M in Series C capital. March Capital led the round, joined by General Catalyst Partners, Insight Partners, E14 Fund, and the American Dental Association. The firm was valued at $55M.
- AI-powered digital health platform Healthee closed its Series A funding round at $32M. Fin Capital, Glilot Capital Partners, and Group11 co-led the funding round, with additional support from TriNet.
For the latest on AI startup funding and more in the VC community, subscribe to our Inside Venture Capital newsletter. | | | |
| 7 | | QUICK HITS - Sales teams at MongoDB, Warmly, and User Evidence are seeing 56% more pipeline from this simple framework.*
- Plagiarism detector Turnitin confirmed a modest layoff of 15 employees after its CEO warned that it could cut staff by 20% due to AI.
- Two Nvidia board members are capitalizing on the stock's surge by offloading shares this week. Directors Mark Stevens and Tench Coxe sold 12,000 and 200,000 shares, respectively, earning them roughly $10.2M and $170M.
- Researchers from Carnegie Mellon and Los Alamos National Laboratory developed an ML model that simulates reactive processes in organic materials, requiring less time and computing power than traditional models.
*This is a sponsored post. | | | |
| Upcoming Events | | MAR
7 | | Portfolio Diversification - Learn from knowledgable experts on the latest trends in alternative investment assets like private equity, crypto, real estate, and the art market. | | | | | | | MAR
27 | | Discover how generative AI can speed up your enterprise, expand your business model, and improve your value proposition. Network with hundreds of AI leaders.* | | | | | | * This is a sponsored event | | | |
 | AI and technology writer | | Beth is a contributing editor and writer of Inside's AI and Tech newsletters. She has written for publications including USA Today, the Arizona Business Gazette, and The Arizona Republic, where she received recognition with a Pulitzer Prize nomination and a First Amendment Award for collaborative reporting on state pension cost increases. You can reach her at Beth.Duckett@yahoo.com. | | This newsletter was edited by Aaron Crutchfield | |
| |
| |
Comments
Post a Comment