• The AI Pulse
  • Posts
  • šŸ¤– OpenAIā€™s New Safety and Security Committee?

šŸ¤– OpenAIā€™s New Safety and Security Committee?

PLUS: Bilingual AI Brain Implant, LLMs Stock-Picking Prowess

Welcome back AI enthusiasts!

In todayā€™s AI Report:

  • šŸš§OpenAIā€™s New Safety and Security Committee?

  • šŸ¤ÆBilingual AI Brain Implant

  • šŸ’µLLMs Stock-Picking Prowess

  • šŸ› 5 Trending Tools

  • šŸ’°Venture Capital Updates

  • šŸ’¼Whoā€™s Hiring?

Read Time: 3 minutes

šŸ—žRECENT NEWS

OPENAI

šŸš§OpenAIā€™s New Safety and Security Committee?

Image Source: Benedikt von Loebell/World Economic Forum

OpenAI has formed a new Safety and Security Committee to make recommendations to the Board of Directors on critical safety and security decisions for AI projects.

Key Details:
  • The Safety and Security Committeeā€™s first task will be to evaluate and further develop OpenAIā€™s processes and safeguards over the next 90 days.

  • The Safety and Security Committee will be led by chairperson Bret Taylor, board member Adam Dā€™Angelo, and OpenAI CEO Sam Altman.

  • The other committee members include Head of Preparedness Aleksander Madry, Head of Safety Systems Lilian Weng, Head of Alignment Science John Schulman, and Head of Security Matt Knight.

  • Former OpenAI board member Helen Toner recently explained that a lack of trust was the primary reason for Altmanā€™s brief firing last year: ā€œWe learned about ChatGPT on Twitter.ā€

  • In October 2023, Toner co-wrote a paper on AI safety and the negative impacts of OpenAIā€™s ChatGPT launch. Altman confronted Toner, stating she harmed the startup by criticizing it publicly.

Why Itā€™s Important:
  • Over the past few months, several of OpenAIā€™s high-profile departures have been from the safety side of its technical teams, including Superalignment co-leader Jan Leike, who believed the startup ā€œwasnā€™t on the trajectoryā€ to get issues of AI security and safety ā€œright.ā€

  • As OpenAIā€™s CEO, Altman is naturally invested in the startupā€™s success. Leading the Safety and Security Committee creates incentive misalignment, as Altman will always choose the option that doesnā€™t hinder OpenAIā€™s growth.

šŸ©ŗ PULSE CHECK

Does Altmanā€™s leadership role on the Safety and Security Committee create a conflict of interest?

Vote Below to View Live Results

Login or Subscribe to participate in polls.

AI-DRIVEN BRAIN IMPLANTS

šŸ¤ÆBilingual AI Brain Implant

Image Source: Canva AI Image Generator

Scientists at the Center for Neural Engineering and Prostheses (CNEP) developed a bilingual brain implant that uses AI frameworks to help a stroke survivor communicate.

Key Details:
  • In 2003, a 20-year-old farmer named Pancho was in a car crash and received emergency surgery to repair his damaged stomach. After a stroke, he became severely paralyzed and could only communicate through moans and grunts.

  • Under the leadership of neurosurgeon Dr. Edward Chan, CNEP deployed a neural implant with neural networks (i.e., AI frameworks designed to mimic the human brain) to decode words based on the brain activity produced when Pancho attempted to articulate them.

  • CNEP scientists discovered Panchoā€™s brain had ā€œcortical activityā€ across English and Spanish, allowing him to ā€œparticipate in a conversation, switching between both languages based on preference.ā€

Why Itā€™s Important:
  • CNEP scientists identified that the brain section responsible for speech (i.e., the Broca area) worked similarly regardless of the language. Instead, brain activity reflected the specific mouth and throat shapes (i.e., vocal tract movements) required to make the sounds.

  • Simply put, the brain cares more about what you want to say than the specific language you use to say it.

AI RESEARCH

šŸ’µLLMs Stock-Picking Prowess

The University of Chicago Booth School of Business investigated whether a Large Language Model (LLM) could analyze financial statements similar to those of a professional human analyst.

They provided standardized and anonymous financial statements to OpenAIā€™s GPT-4, instructing the AI model to analyze them and determine the direction of future earnings.

Despite not having access to narrative or industry-specific information, OpenAIā€™s GPT-4 outperformed professional human analysts at predicting future earning trends.

When leveraging chain-of-thought prompting (i.e., guides LLMs to solve problems through step-by-step reasoning) to emulate human reasoning, OpenAIā€™s GPT-4 achieved a future earning accuracy of 60%, which is remarkably higher than that of a professional human analyst.

The AI modelā€™s ability to analyze financial statements came from its real-time analysis of trends, ratios, and economic reasoning without relying on memorizing patterns in training data.

šŸ› TRENDING TOOLS

āš™ļøFrontly builds AI-powered apps with no code.

šŸŽøFlowTunes generates endless chill lo-fi beats to focus.

šŸ“ŠRoadway takes the guesswork out of growth marketing.

šŸ§ ReplyMind crafts relevant, thoughtful, and funny replies with just a click.

āœļøMetaview generates AI-powered notes for debriefs, interviews, and screening calls.

šŸ”®Browse our always Up-To-Date AI Tools Database.

šŸ’°VENTURE CAPITAL UPDATES

šŸ’¼WHOā€™S HIRING?

  • Onsemi (San Jose, CA): Python Developer Intern, Fall 2024

  • AMD (Santa Clara, CA): Graduate Hardware Engineering Co-Op/Intern, Fall 2024

  • Genetech (San Francisco, CA): PhD Research Intern, Foundational ML and AI, Fall 2024

  • Motorola Solutions (Somerville, MA): Machine Learning (ML) Engineer Co-Op, Fall 2024

  • Applied Intuition (Mountain View, CA): Technical Recruiter, New Grad

šŸ¤–PROMPT OF THE DAY

MARKETING

šŸŒ‚Brainstorm Marketing Channels

Envision yourself as a marketing expert. Recommend the most effective marketing channels for [Business] with [Product/Service], [Target Audience], and [Budget].

Business = [Insert Here]

Product/Service = [Insert Here]

Target Audience = [Insert Here]

Budget = [Insert Here]

šŸ“’FINAL NOTE

If you found this useful, follow us on Twitter or provide honest feedback below. It helps us improve our content.

How was todayā€™s newsletter?

ā¤ļøTAIP Review of the Day

ā€œI feel like The AI Pulse is getting better and better with every read.ā€

-Ali (ā­ļøā­ļøā­ļøā­ļøā­ļøNailed it!)
REFER & EARN

šŸŽ‰Your Friends Learn, You Earn!

You currently have 0 referrals, only 1 away from receiving āš™ļøUltimate Prompt Engineering Guide.

Refer 9 friends to enter šŸŽ°Mayā€™s $200 Gift Card Giveaway.

Reply

or to participate.