| |
| Here’s a look at today's AI briefing: - Microsoft to open AI hub in London.
- New York Times reports on AI companies' controversial data collection methods.
- Spotify debuts AI-generated playlists.
- Musk predicts AI 'smarter than any one human' at end of 2025.
- AWS Survey: 86% of European employers to use AI by 2028.
Beth
p/beth-duckett | |
| 1 | | Microsoft is launching an AI hub in London, to be led by AI scientist and engineer Jordan Hoffmann. DeeMind co-founder Mustafa Suleyman, recently hired by Microsoft, announced the new hub in a blog post. More: - Suleyman was hired to lead Microsoft's new unit focused on consumer AI products and research.
- That unit, Microsoft AI, will open a hub "in the heart of London" to support advanced LLMs, infrastructure, and tools for foundation models, he wrote.
- The team will work "closely with our AI teams across Microsoft and with our partners, including OpenAI," Suleyman added.
- Microsoft plans to tap into the U.K.'s large pool of AI talent and post job openings in the coming months.
Zoom out: - In November, Microsoft said it plans to invest £2.5B ($3.2B) in expanding its next-gen AI data center infrastructure in the U.K. over the next three years.
|     | |
| 2 | | The New York Times reports that OpenAI, Google, and Meta have potentially engaged in questionable practices to train their AI models, including using copyrighted works and transcribing YouTube videos for new data. The report comes as AI companies face challenges in acquiring high-quality training data. More: - According to the Times, OpenAI used its Whisper speech recognition tool to transcribe over a million hours of YouTube videos.
- The effort to train GPT-4 continued despite the company's legal concerns and potential violations of YouTube's rules.
- An OpenAI spokesperson told The Verge that the company uses public data and partnerships for private data, and is exploring synthetic data.
Meta and Google: - Google also created YouTube transcripts and had its legal department quietly change policy language to broaden its use of consumer data.
- Meta reversed its stance on using copyrighted material for AI training after noting OpenAI's actions, per the Times.
- The report suggests Meta considered acquiring publisher Simon & Schuster LLC to secure works for AI training.
Zoom out: - Despite facing lawsuits claiming intellectual property theft, Google, OpenAI, and Meta continue to use copyrighted data for AI without waiting for legal clarity.
- The New York Times sued Microsoft and OpenAI in December, seeking billions of dollars in damages due to alleged "unlawful copying" of its news stories to train GPT models.
- In a blog response, OpenAI argued that no single source, including The Times, significantly contributes to the training of its models.
| | | |
| A message from our sponsor, SocketLabs. |  | | Ask Google and Yahoo Your Most Pressing Email Marketing Questions! This year, Google and Yahoo rolled out new rules for email marketers to reduce spam and create a better experience for their users. Are your emails still making it to the inbox? On April 10 at 12 p.m. EST, SocketLabs will host a live and interactive discussion with Dan Givol, group product manager at Google, and Clea Moore, principal product manager at Yahoo, to discuss bulk sender requirements, how to know if you’re running into compliance issues, and why these changes are designed to make your email campaigns more successful. Moderated by SocketLabs’ CMO Lauren Meyer, attendees will: - Hear more about Google and Yahoo sender requirements and why they’re in place
- Get email marketing recommendations for better, more reliable inbox placement
- Ask their most important email questions
Join us for this exclusive opportunity to talk email with THE email experts! Register Here | |
|
| 3 | | Spotify said its paid users can now create AI-generated playlists from text prompts. The beta feature is first launching to Premium subscribers in the U.K. and Australia. More: - Spotify said it's using LLMs and personalization technology based on user history and preferences to create the customized AI playlists
- Users enter the prompts into a text field on the Spotify mobile app for iOS or Android.
- Spotify's announcement gave example prompts, such as "relaxing music to tide me over during allergy season" and "a playlist that makes me feel like the main character."
- Its AI tailors the playlists over time to user preferences. Users can also edit the lists with additional prompts.
Zoom out: - Spotify launched an AI-powered "DJ" last year. The feature curates music based on a user's listening history and adds synthetic voice commentary.
| | | |
| 4 | | Elon Musk predicts AI could surpass the smartest human within the next one to two years. Musk made remarks about AI during an interview today with Norway wealth fund CEO Nicolai Tangen. More: - Musk said he predicts there will be AI "smarter than any one human" probably by the end of 2025.
- He added that the total "sentient compute of AI" could surpass that of all humans within five years.
- AI's advancement, however, remains limited by the availability of electricity, he noted.
Zoom out: - Musk also said his AI startup, xAI, should finish training a new Grok AI model by May.
- The updated Grok "should be better than GPT-4," followed by model orders of magnitude more powerful, he said.
- Musk stressed the need for AI regulation but warned that its rapid development could outpace oversight. He also discussed risks of programming AI for political correctness.
| | | |
| A message from our sponsor, Lacework. | | 4/18 webinar: Compromised credentials in 2024? According to Lastpass, professionals manage up to 200 sets of credentials on average. When credentials are compromised, cyber attackers gain frictionless entry into sensitive systems and can often move laterally to find your crown jewels. Attend this webinar to understand: - Execution methods behind compromised credential attacks
- What the bad actors do with stolen identities
- Preventative best practices to implement today
Register now for Lacework's deep dive on what cybersecurity professionals need to know about the world's #1 attack vector. Register Now | |
|
| 5 | | In 2023, just over 70% of European employers used AI tools at work, with 86% planning to do so by 2028, an AWS survey shows. 
What happened: Access Partnership and AWS surveyed over 6,500 employees and 2,000 employers across France, Germany, Spain, and the U.K. What the numbers show: Spain is predicted to lead in AI workplace adoption by 2028, with 92% of employers expecting to use AI tools, while the U.K. anticipates the biggest increase in AI use from 67% to 85%. What else: The survey also found that nearly 80% of employees plan to use generative AI tools within 5 years. By 2028, generative AI is forecasted to be a core part of many workplaces, with 91% of employers and 86% of employees seeing its benefits. Most European employees view AI positively for career growth, with 65% eager to learn AI skills for advancement and potentially earning 27% in higher wages. | | | |
| 6 | | AI Events and Conferences: - World Summit AI Americas: April 24-25, Montreal.
- Data Innovation Summit: April 24-25, Stockholm and online.
- Rise of AI Conference: May 14-15, Berlin, Germany, and virtual.
- Enterprise Generative AI Summit West Coast: May 21-22, Silicon Valley, California.
- AI Expo Tokyo: May 22-24, Tokyo.
- Data Cloud Summit 2024: June 3-6, San Francisco.
- SuperAI: June 5-6, Singapore.
- The AI Summit London: June 12-13, London, U.K.
- 2024 IEEE Conference on Artificial Intelligence: June 25-27, Singapore.
- Ai4 2024: Aug. 12-14, Las Vegas, Nevada.
- The AI Conference: Sept. 18-19, San Francisco.
| | | |
| 7 | | QUICK HITS: - When it's time to grow your company, make Ohio your competitive advantage with better sites, sharper talent, earlier funding, and less red tape.*
- Canada's government will allocate C$2.4B ($1.8B) in its upcoming budget to bolster the country's AI sector, including investments in computing infrastructure.
- Alibaba is offering discounts on its cloud services globally as demand for AI projects rises.
- The VC arm of Italy's CDP will invest $1.1B in AI and cybersecurity over five years.
- Elon Musk said Tesla will reveal its robotaxi on Aug. 8. The next-gen electric vehicle could be designed entirely for self-driving, with no pedals or steering wheel.
*This is a sponsored post | | | |
| Upcoming Events | | APR
10 | | Ask email experts from Google and Yahoo your most important questions at SocketLabs' interactive webinar on April 10. | | | | | | * 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 Beth Duckett | |
| |
 | Ask email experts from Google and Yahoo your most important questions at SocketLabs’ interactive webinar. | |
|
 | Lacework is a data-driven security platform for the cloud and the leading cloud-native application protection platform solution. | |
|
| |
Comments
Post a Comment