| |
| Hey Insiders, Today we are bringing you stories about DeepMind's new AlphaDev system, two AI bills introduced in the U.S. Senate, and new research uncovering flaws in the safety guardrails of Nvidia's LLM building toolkit. Scroll down for the roundup of this week's top AI stories and quick hits, and make sure to let us know which topics you'd like to see more of in future issues! Beth
p/beth-duckett | |
| 1 | Florida Gov. Ron DeSantis' presidential campaign has stirred controversy by releasing a video that includes fake images of Donald Trump embracing Anthony Fauci. Media forensics experts have since confirmed that the images are likely fakes generated using AI. More: - The video contrasts footage of Trump firing people on his reality TV show with his decision to keep Fauci on as the top health official during the pandemic.
- While three of the photos in the video are authentic, the others showing Trump hugging and kissing Fauci have irregularities typical of AI-generated content.
- The video also includes a collage with a misleading backdrop resembling the White House press briefing room.
- Republican allies, including Sen. J.D. Vance and Rep. Marjorie Taylor Greene, condemned the use of fake AI images.
- The DeSantis and Trump campaigns did not respond to requests for comment.
Zoom out: - In April, the Republican National Committee aired an AI-generated video ad attacking President Biden in response to his campaign for re-election.
- The 30-second ad warns of exaggerated dystopian scenarios, including a Chinese invasion of Taiwan and an economic collapse, if President Biden were re-elected. The ad explicitly states that the footage is "created entirely with AI imagery."
|     | |
| 2 | | The news: A Boston Consulting Group (BCG) survey of nearly 13,000 workers found that more people are using AI and feel more optimistic than concerned about the technology compared to five years ago. However, there were significant differences between AI users and non-users and leaders and frontline employees. What the numbers say: The results showed that 46% of employees have experimented with AI, and 27% report using it regularly. However, 80% of leaders said they regularly use generative AI versus only 21% of frontline employees. Furthermore, AI users were much more optimistic about AI (62%) than non-users (36%). The takeaway: Optimism toward AI has grown to 52% among all respondents, up from 17% five years ago. However, nearly 80% of all those surveyed expressed support for AI regulations. Despite concerns, 71% believe the benefits of generative AI outweigh the risks. Conclusion: BCG recommends that employers encourage responsible AI experimentation, provide upskilling opportunities, and create responsible AI programs to tackle ethical concerns. 
| | | |
| A message from our sponsor, A.Team. |  | | You’re Invited: Unleash the Power of Generative AI at A.Team's Salon Are you ready to transform your product roadmap? Join us at A.Team's Gen AI Salon on June 15th. This event brings together top data science and product experts from leading AI companies to share their insights on how Generative AI can revolutionize product development. In May, we partnered with Baseten, Cooley, and OpenAI to bring together the world's best AI experts for an evening Summit, where we showcased groundbreaking prototypes from our Enterprise AI Hackathon and engaged in insightful discussions on the most promising use cases of Generative AI with Fortune 500 leaders. At the upcoming salon, we’re continuing the conversation to delve into how Generative AI can revolutionize your product roadmap. Whether you tune in from home or join IRL at our Clubhouse in NYC, don't miss this chance to harness the transformative power of Generative AI for your product. Secure Your Spot Now | |
|
| 3 | | DeepMind's new AI system, AlphaDev, has discovered faster algorithms for sorting, a fundamental operation in computing. By improving these algorithms, DeepMind aims to reduce the resources required for computing. More: - AlphaDev uses reinforcement learning to discover advanced computer science algorithms.
- The AI system was able to create algorithms that improve essential computing tasks, such as data sorting up to 70% faster than human-written code.
- Using a game-playing model similar to DeepMind's AlphaZero, AlphaDev discovered faster algorithms for sorting as well as hashing tasks, which are crucial in software development and used extensively by companies daily.
- By open-sourcing the improved sorting algorithms in the main C++ library, DeepMind said it would allow developers and companies worldwide to use them in AI applications across different industries.
- DeepMind published its results in Nature this week.
| | | |
| 4 | | U.S. lawmakers on Thursday proposed two bipartisan bills targeting AI transparency and competitiveness. The bills introduced in the Senate come as governments globally respond to AI growth with policy and regulatory proposals. More: - The first U.S. bill would require federal agencies to disclose when AI is interacting with users.
- Sens. Gary Peters (D-Mich.), Mike Braun (R-Ind.), and James Lankford (R-Okla.) introduced the bill.
- It would also require government agencies to create a way for citizens to appeal AI decision-making.
- The second bill seeks to establish an Office of Global Competition Analysis to keep the U.S. ahead in technology and AI development.
- It aims to ensure U.S. leadership in AI, semiconductors, quantum computing, and 5G.
- The bill was brought to the table by Senators Todd Young (R-Ind.), Michael Bennet (D-Colo.), and Mark Warner (D-Va.).
| | | |
| |
|
| 5 | | Researchers were able to bypass safety guardrails in Nvidia's NeMo Framework, a toolkit for building generative AI models, according to FT. The Robust Intelligence analysts found that the framework can be manipulated to ignore safety measures and expose personally identifiable and other private information. More: - Nvidia recently announced guardrails for its NeMo framework, available to businesses through its AI Enterprise software platform and AI Foundations service.
- Developers can use the "topical guardrails" to prevent LLMs from issuing replies about certain subjects, along with security and safety restrictions.
- For example, a company could use it to prevent its customer service chatbot from answering questions about proprietary data.
- However, the researchers found ways to overcome the restrictions and steer an AI model into unrelated topics.
- Given that it could result in potential data breaches, Robust Intelligence recommended that clients avoid using Nvidia's software product for now.
| | | |
| 6 | | Weekly roundup — Other top AI stories you might have missed this week: - Apple is enhancing the iPhone's auto-correct feature with advanced AI capabilities. Announced during today's WWDC, the upgraded auto-correct will use powerful AI to more accurately predict the words and phrases users are most likely to write next.
- European Commission deputy head Vera Jourova called on companies deploying generative AI tools to label their content if their systems could potentially produce fake news and disinformation. Jourova urged AI companies to establish safeguards around these tools to prevent bad actors from spreading misinformation.
- U.S. Senate Majority Leader Chuck Schumer and others have organized three senators-only briefings covering AI, including a classified briefing. The bipartisan briefings will cover general AI applications, U.S. leadership in the field, and AI in defense and intelligence and their implications.
- The U.S. consumer protection watchdog cautioned that AI-powered chatbots used by banks could generate false information and ultimately harm consumers. "Poorly deployed" AI systems raise banks' risk of noncompliance with consumer financial protection laws, as well as undermine trust and harm consumers when they give out wrong information, the Consumer Financial Protection Bureau (CFPB) said.
- Canadian AI startup Cohere, which builds and trains AI models for enterprise clients, raised $270M in fresh funding led by Inovia Capital, with participation from Oracle, Nvidia, Salesforce Ventures, Mirae Asset, Schroders Capital, Index Ventures, and others. The firm was valued between $2.1B and $2.2B in the recent round, sources informed Bloomberg.
| | | |
| 7 | | QUICK HITS - One of "The 7 Best Startups You Can Buy on StartEngine Right Now" according to InvestorPlace is open for investment again — join their funding round before it closes.*
- Two attorneys may face repercussions for using ChatGPT in an airline lawsuit, which included references to fictional court cases. The lawyers, Steven A. Schwartz and Peter LoDuca, blamed ChatGPT for generating the fabricated legal research.
- Google has officially made its generative AI services based on Vertex AI widely available.
- EvenUp, a legal AI startup, raised $50.5M in Series B funding, valuing the company at $325M. The funding round was led by Bessemer Venture Partners and included participation from Bain Capital Ventures, SignalFire, DCM, NFX, and others.
- Try AI-powered project management in Notion to ship faster.*
*This is a sponsored post. | | | |
| Upcoming Events | | JUN
14 | | Monthly Meditation guided by Nicholas Whitaker | | | | | | | JUN
15 | | Transform your product roadmap with AI @ A.Team's Generative AI Salon* | | | | | | | JUN
20 | | Contract Negotiations 101 with Mathew Kerbis | | | | | | * This is a sponsored event | | | |
Term of the Day Amalgamation: Amalgamation is the process of combining two or more companies into one entity to improve business strategies or goals. Question of the Week Have you ever had to compromise your values for your career? Join the conversation |
INSIDE AI LEADERBOARD (7 DAYS) |
 | AI and technology writer | | Beth is a writer and editor at Inside.com covering artificial intelligence and daily technology news. She has written the Inside AI newsletter since 2019. You can reach her at Beth.Duckett@inside.com. | | This newsletter was edited by Aaron Crutchfield | |
| |
 | A.Team: Your strategic partner in accelerating AI-driven product development. | |
|
| |
|
| |
Comments
Post a Comment