| |
| Here's a look at today's AI briefing: - Comedy special features AI-generated George Carlin.
- Snapchat allows parents to block teens' access to AI chatbot.
- Study finds popular LLMs frequently generate false legal information.
- Rabbit sells out of AI-powered assistant device in a day.
Beth
p/beth-duckett | |
| 1 | | An AI comedy platform has brought back the late stand-up comedian George Carlin through an AI-generated special called "George Carlin: I'm Glad I'm Dead." The digital recreation of Carlin did not receive permission from his family. His daughter, Kelly Carlin, denounced the project, stating that "No machine will ever replace his genius." More: - The one-hour special was produced by Dudesy, a podcast hosted by "Mad TV" alum Will Sasso and podcaster Chad Kultgen that features an AI program, also called Dudesy.
- The AI comedian, which was trained on Carlin's material, addresses topics including U.S. gun culture, transgender issues, Taylor Swift, and the revolutionary potential of AI.
- Shortly after the special debuted online, Carlin's daughter publicly criticized it on X.
- "These AI-generated products are clever attempts at trying to recreate a mind that will never exist again," she wrote. "Let's let the artist's work speak for itself. Humans are so afraid of the void that we can't let what has fallen into it stay there."
- Dudesy's first comedy special, featuring an AI-generated voice of quarterback Tom Brady, was removed due to legal threats from Brady.
|     | |
| 2 | | Snapchat is adding parental controls to let parents block their teens from using the app's AI chatbot. If parents choose to deactivate the bot, teens can still message "My AI," but it will reply with a message indicating that it's been turned off More: - The chatbot, called My AI, already underwent improvements last year after tests revealed inappropriate responses to messages from minors,
- The bot, based on OpenAI models, has received backlash for engaging with minors on sensitive topics without age restrictions.
- Since launching nearly a year ago, Snapchat users have shared their interesting and sometimes strange conversations with the bot.
- Snap has said the GPT-powered chatbot is customized to generate text specifically for Snapchat, with a "unique tone and personality" geared toward the platform's values on "friendship, learning, and fun."
Zoom out: - Last year, a U.K. watchdog raised concerns regarding children's privacy risks associated with Snap's AI chatbot.
- The bot also autonomously shared its own self-generated video on the platform, triggering confusion and even fear among some users.
- Snap claims its AI chatbot underwent a comprehensive legal and privacy review before launching to the public.
| | | |
| A message from our sponsor, Calendly. |  | | Turn scheduling from a pain point into a competitive advantage The average meeting takes 7.5 emails to schedule. All that back-and-forth, plus reminders and follow-ups — it's a poor experience for you, your customers, and your prospects. Calendly lets you eliminate that busywork and delight everyone you meet. Teams that use Calendly are seeing: - 2x conversion rates (CallRail)
- +30% customer retention (Conductor)
- 75% faster time-to-interview (Base)
- 323% ROI (CI Assante Wealth Management)
How are teams getting these results? Here are 4 features Calendly users can't live without: - Embed: Schedule directly from your website.
- Routing: Qualify & route meetings to the right person.
- Workflows: Automate reminders, follow-ups & more.
- Integrations: Integrate with Salesforce, Hubspot, Zoom, & 100s more.
Sign up for free to join 20 million people and 100k companies using Calendly to power their scheduling. Sign Up Free | |
|
| 3 | | The most popular generative AI models often generate false legal information, or "hallucinations," when asked questions about federal court cases, according to a new study from Stanford RegLab and Institute for Human-Centered AI researchers. In the pre-print study, hallucination rates for state-of-the-art language models answering legal questions varied between 69% and 88%. 
By the numbers: According to the study, GPT-3.5 had a 69% hallucination rate, Google's PaLM 2 had 72% false responses, and Meta's Llama 2 gave false information 88% of the time. The researchers described the legal hallucinations as "alarmingly prevalent," adding that the models did worse when asked more complex questions. The models also lacked self-awareness about their errors and tended to reinforce incorrect assumptions and overstate their confidence in responses, they noted. Why it matters: The findings suggest that employees in the legal field take a "cautious and supervised" approach when using generative AI models for work. The field has already seen the repercussions of these hallucinations: In June, a judge fined two Manhattan attorneys $5,000 for filing a legal brief that included fabricated court cases and citations generated by ChatGPT. Michael Cohen, Donald Trump's former attorney, also mistakenly used fake legal citations from Google's AI tool Bard in a brief in November. Cohen admitted that he thought Google Bard was a "supercharged search engine," not realizing it generates text like ChatGPT and can create citations and descriptions that appear "real but actually were not." | | | |
| 4 | | Representatives from OpenAI, Anthropic, and Cohere secretly met with Chinese AI experts in Geneva to discuss AI's potential to spread misinformation and disrupt social cohesion, sources told The Financial Times. Scientists and policy experts from the AI companies reportedly met with representatives from Tsinghua University and other Chinese state-backed institutions last year. More: - The White House, U.K., and Chinese government officials were aware of the Geneva meetings before they took place in July and October of 2023.
- The Shaikh Group, a private mediation organization, arranged the talks.
- Both sides discussed AI's risks, policy proposals, areas to collaborate in technical cooperation, and AI safety research investments, according to FT.
- The goal is to uncover a safer scientific route for advancing AI technologies.
- According to a negotiator, the successful meetings have paved the way for future talks focused on aligning AI systems with legal codes and societal norms.
| | | |
| A message from our sponsor, Placer.ai. |  | | 8 Retail Trends to Watch in 2024 2023 was another tumultuous year as inflation, high interest rates, and general economic headwinds continued to constrain consumer behavior. Still, this year also revealed many bright spots across various consumer-facing sectors, from retail to dining to advertisement to entertainment. Which of these developments are likely to shape the consumer landscape in 2024? Read the whitepaper to find out: - Why will events continue creating significant opportunities for their surrounding areas?
- How will placemaking impact everything from shopping center design to office space layout to retailer's product offerings?
- Where will dining and retail concepts focus their expansion efforts?
Read the white paper | |
|
| 5 | | The Rabbit R1, a pocket-sized AI gadget designed to control a user's apps, quickly sold out of its first batch, with the startup announcing 10,000 units sold in a day. While the first batch will ship in late March, Rabbit has made a second batch of R1s available for pre-order, with shipments in April or May. More: - Designed by Teenage Engineering, the R1 can control a phone's apps by using its push-to-talk button, similar to a walkie-talkie.
- The square-shaped device, about half the size of an iPhone, can also answer questions in natural language.
- It features a 2.88-inch touchscreen, two microphones, a speaker, an analog scroll wheel, and a camera that can be rotated.
- Rabbit's major innovation, its Large Action Model (LAM), powers its operating system to perform different tasks through user interfaces rather than apps or APIs.
- Prices start at $199.
| | | |
| 6 | | Microsoft briefly surpassed Apple as the most valuable U.S. company on Thursday. Analysts credited Microsoft's rapid growth and leadership in the generative AI revolution for the latest surge. More: - Apple's shares have fallen 4% in 2024 amid concerns about weakening smartphone demand, while Microsoft has gained 2% this year.
- Microsoft has briefly overtaken Apple as the most valuable company several times since 2018, including during iPhone supply chain challenges in 2021.
- Apple, meanwhile, is facing the possibility of weakening demand for iPhones and Macs.
- Apple iPhone sales in China plummeted by 30% in the first week of 2024, primarily due to competition from local rivals such as Huawei, according to Jefferies analysts.
| | | |
| 7 | | QUICK HITS - Develop AI tools from the world's best models, leverage any data source, and deploy for any internal or external use with MindStudio's no-code platform. Get Started.*
- OpenAI CEO Sam Altman and U.S. House Speaker Mike Johnson (R-La.) met on Capitol Hill today to discuss the potential dangers of AI.
- A bipartisan group of U.S. lawmakers unveiled a bill that would require federal agencies and their AI vendors to follow best practices for mitigating AI-related risks.
- Indian data center operator Yotta will buy an additional $500M worth of AI chips from Nvidia, raising its total order with the U.S. firm to $1B.
- Google Cloud announced new generative AI-powered tools for retailers, including a chatbot that provides product recommendations to customers and an LLM that improves product searches.
- Smart investors are generating passive income by selling their shareholder voting rights. Register on Shareholder Vote Exchange to start earning.*
*This is a sponsored post. | | | |
| Upcoming Events | | JAN
30 | | Join TruEra for an expert discussion about the impact of the EU AI Act on US and Canadian firms | | | | | | | MAR
7 | | Alternative investment assets to know in 2024. | | | | | | * 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 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 | |
| |
 | Calendly is your scheduling automation platform for eliminating the back-and-forth — and much more. | |
|
| |
|
| |
Comments
Post a Comment