| |
| Here's a look at today's top AI stories: - Microsoft announces custom-designed AI and cloud chips.
- U.S. senators introduce bill for AI transparency and accountability.
- Survey shows U.S. Christians less hopeful about AI's impact than non-Christians.
- Forward raises $100M to deploy AI-powered healthcare kiosks.
Thanks for reading! Beth
p/beth-duckett | |
| 1 | | Microsoft unveiled its first custom AI chip for powering AI workloads in the cloud. Announced during Microsoft's Ignite event today, the Azure Maia 100 AI Accelerator was designed to run large language models (LLMs) and speed up AI computing tasks. More: - Maia is tailored for handling cloud-based AI workloads, such as training and inference for LLMs.
- Microsoft will not sell the Maia chips. Instead, the company will deploy the AI accelerators at its data centers to power its chatbot, "Copilot" service, and Azure OpenAI cloud computing service.
- The company is already testing the chip for its Bing and Office AI products, as is its partner, OpenAI.
- Microsoft reportedly designed the Maia chip to be specifically optimized for OpenAI's GPT.
Zoom out: - Also on Wednesday, Microsoft debuted its own Arm-based central processing unit, Azure Cobalt, for cloud workloads.
- Microsoft also plans to provide Azure customers with cloud services running on the latest Nvidia and AMD AI chips starting next year.
|     | |
| 2 | | Here are the other AI-related announcements at Microsoft's Ignite conference: - Microsoft unveiled Windows AI Studio, a central hub for developers to access and customize AI models. The platform integrates development tools and models from Azure AI Studio and services like Hugging Face.
- Microsoft Teams is getting updates, including AI-powered noise reduction and a "background decoration" feature. The latter will use generative AI to declutter and improve the look of video call backgrounds.
- Microsoft has rebranded its Bing chat interface to "Microsoft Copilot," transitioning away from the branding across Bing, Microsoft Edge, and Windows 11. Bing Chat and Bing Chat Enterprise are now simply named "Copilot."
- OpenAI's new GPT-4 Turbo model will debut in preview this week in Azure OpenAI. GPT-4 Turbo, trained with data up to April 2023, has cheaper pricing than GPT-4 and an expanded context window.
- Microsoft will also support OpenAI's newly launched GPTs. The custom versions of ChatGPT can be created by anyone for personal and work tasks.
Zoom out: - Also at Ignite, Nvidia announced it's bringing an AI foundry to Microsoft's Azure cloud, allowing customers to create and fine-tune customized generative AI models.
- During the conference, Nvidia CEO Jensen Huang called GenAI "the single most significant platform transition in computing history," saying it will be bigger than PCs, mobile, and the internet.
| | | |
| A message from our sponsor, Okta. |  | | Empower Your Small Business with Okta's Secure Identity Solutions Unlock the full potential of your small business with Okta's cutting-edge identity management solutions. In today's digital landscape, security and efficiency are key, and Okta delivers on both fronts. Our robust platform offers seamless authentication, single sign-on, and multi-factor authentication to safeguard your business data. But it's not just about security; it's also about efficiency. Okta streamlines user onboarding, making it a breeze to manage access to your applications and resources. Whether you're scaling up or streamlining operations, Okta adapts to your needs. Join the thousands of small businesses that trust Okta to protect their digital assets while enhancing productivity. Get started today by visiting our Okta for Small Business landing page. Your business deserves the best in identity management. Learn More 
| |
|
| 3 | | U.S. Sens. Amy Klobuchar (D-Minn.) and John Thune (R-S.D.) introduced a bill that seeks to establish standards for AI transparency and accountability. The bipartisan Artificial Intelligence Research, Innovation, and Accountability Act of 2023 would require "critical-impact" AI organizations to seek certification for their AI systems, among other directives. More: - The bill would direct NIST, the National Institute of Standards and Technology, to develop risk-based recommendations for "high-impact" AI systems.
- Companies using these AI systems would have to turn in transparency reports to the U.S. Commerce Department.
- The Department would be required to submit and periodically update a five-year plan for testing and certifying the "critical-impact" AI.
- The bill also defines certain AI terms and would require consumers to be notified about generative AI images.
- It would direct NIST to standardize AI image detection methods.
Zoom out: - Thune said the legislation would bolster U.S. innovation and leadership in AI "while also establishing common-sense safety and security guardrails for the highest-risk AI applications."
- The bill's co-sponsors include Roger Wicker (R-Mich.), John Hickenlooper (D-Colo.), Shelley Moore Capito (R-W.Va.), and Ben Ray Luján (D-N.M.), who are all members of the Commerce Committee.
| | | |
| 4 | | U.S. Christians and non-Christians hold differing opinions on the positive traits of AI, according to a survey of 1,500 U.S. adults by Barna and Gloo. About 28% of Christians expressed hope that AI can have a positive impact, whereas 39% of non-Christians shared the same view. Christians also tend to be less fascinated or excited about AI and its potential to make life easier. 
What the numbers show: Only 20% of U.S. Christians said they agree that AI is strongly or somewhat beneficial for the Christian Church. Just over 50% said they disagree that it's beneficial. Meanwhile, 52% of Christians polled said they would be disappointed if they discovered that their church was using AI. The bigger picture: In its analysis of the survey, Barna, a non-partisan for-profit organization, cited Kenny Jahng, founder of AiForChurchLeaders.com and editor-in-chief of ChurchTechToday.com. Jahng said there is a fear that AI will take over and it will be "human versus machine," but a wider perspective shows "there are things that AI is really good for," like brainstorming or learning. "The more constructive way is to think of AI as a super-intelligent student intern," Jahng said. | | | |
| |
|
| 5 | | Health tech startup Forward announced a $100M funding round to deploy AI-powered modules for automated healthcare checkups. The primary care provider is rolling out its "CarePod" modules in the Bay Area, New York, Chicago, and Philadelphia, targeting malls, gyms, and offices. More: - The CarePods, described as standalone medical stations resembling kiosks, can read patients' blood pressure, swab throats, and draw blood.
- A robotic voice directs patients to perform the readings. The company uses large language models to process current healthcare research and pre-generate protocols.
- The custom AI helps reach a diagnosis, with licensed doctors or nurses interpreting the results via video or chat.
- The cost is $99 per month for access to all CarePods. Forward also charges $149/month for virtual and in-person primary care at its 17 U.S. clinics.
Zoom out: - Forward's latest funding round was backed by Khosla Ventures, Founders Fund, Softbank, and others.
- The funding will help Forward initially launch 25 CarePods, with plans to scale up to 3,200 within a year.
- The startup has raised $657M to date.
| | | |
| 6 | UnitedHealth Group faces a lawsuit for allegedly employing a flawed AI algorithm to unfairly deny Medicare patients' rehab care coverage. The proposed class action suit cites a recent STAT investigation that exposed the company's use of the algorithm, which forecasts how long patients should stay in rehab. More: - The suit was brought this week by the California-based Clarkson Law Firm on behalf of the families of two deceased patients with UnitedHealthcare Medicare Advantage plans.
- It alleges that UnitedHealth and its subsidiary, NaviHealth, used the algorithm to "systematically deny" claims for Medicare patients recovering from debilitating illnesses in nursing homes.
- The complaint argues that the AI system had a 90% error rate, wrongly overriding doctors' recommendations.
- UnitedHealth still pushed medical staff to follow the algorithm, with a goal to maintain patients' rehab stays within 1% of the AI's predictions, according to internal documents.
Zoom out: - The class could encompass tens of thousands of patients, with potential damage claims totaling billions of dollars, according to the Clarkson Law Firm.
- UnitedHealth has not commented on the suit. The company has previously claimed that the algorithm was only a guidepost to gauge patient recoveries.
| | | |
| 7 | | Quick Hits: - Elevate your Kubernetes skills with this must-read resource for building cloud-native applications. Get your Kubernetes Cookbook today.*
- Niantic's augmented reality pet game, Peridot, is integrating generative AI to make its digital pets appear more lifelike.
- Generative AI will be used to create deepfakes that will result in "a tsunami of disinformation" ahead of the 2024 U.S. elections, according to Darrell West, senior fellow at the Center for Technology Innovation at the Brookings Institution.
- Chinese tech giant Tencent Holdings said it will seek domestic sources for AI training chips due to recent U.S. restrictions on chip exports to China. In the near term, Tencent has stockpiled enough Nvidia AI chips to continue working on its AI model Hunyuan.
*This is sponsored content. | | | |
| Upcoming Events | | NOV
25 | | AI in Financial Services with Saroop Bharwani | | | | | | | NOV
25 | | The Future of Fractional Work with David Berkowitz | | | | | | * This is a sponsored event | | | |
Term of the Day Post-money valuation: A post-money valuation is a company's equity value after receiving cash from a round of financing. Question of the Week What perks would convince you to return to the office? Join the conversation |
 | AI and technology writer | | Beth is an editor and writer of Inside's AI and Tech newsletters, with a career spanning journalism since 2007. She has written for publications including The Arizona Republic and USA Today and authored a book on the solar industry in 2016. Reach her at Beth.Duckett@yahoo.com. | | This newsletter was edited by Aaron Crutchfield | |
| |
 | Grow your small business simply and securely with Okta. | |
|
 | With 500+ 5-star G2 reviews, you’ll see why 3,500+ customers choose Drata to automate compliance for SOC 2, ISO 27001, GDPR and more. | |
|
| |
Comments
Post a Comment