Welcome, AI RESEARCHERS.

OpenAI's new safety committee, led by Sam Altman and his allies, raises concerns.

With former board members speaking out, a past safety lead joining a rival, and ongoing conflicting commentary, is Sam’s new team a genuine effort or just a PR play? Let's dig deeper and find out.

In today’s AI Researchs:

• OpenAI Launches Safety Committee Amid New AI Model Training

• Finally revealed: Why Sam Altman Was Fired from OpenAI

• AI Outshines Analysts in Financial Forecasting

• Access and Use Google multimodal Gemini 1.5 Pro

• 3 ChatGPT-4o prompts for job search 

• AI chart & AI pointers

• Trending Ai tools

• More Ai hits & news & Researchs

Read time: 4 minutes

OpenAI

🚨 OpenAI Launches Safety Committee Amid New AI Model Training 

AI Researchs: OpenAI has started training its next frontier model, aiming to push the boundaries of AI capabilities towards achieving AGI (Artificial General Intelligence).

For more details:

• OpenAI has initiated training of its next frontier model, targeting advancements towards AGI.

• The newly formed Safety and Security Committee will oversee safety and security decisions, making recommendations within 90 days.

• The committee is chaired by Bret Taylor, with members Adam D’Angelo, Nicole Seligman, and CEO Sam Altman.

• Key experts involved include Aleksander Madry, Lilian Weng, John Schulman, Matt Knight, and Jakub Pachocki.

• External advisors, such as Rob Joyce and John Carlin, will support the committee’s efforts.

Why this is important: This committee's work is crucial for ensuring the safe development and deployment of advanced AI systems, impacting future innovations and industry standards.

Inside the OpenAI Drama

❓ Finally revealed: Why Sam Altman Was Fired from OpenAI.

AI Researchs: Former OpenAI board member Helen Toner recently revealed the reasons behind CEO Sam Altman's firing on The TED AI Show podcast. She highlighted issues of misinformation and lack of trust that led to the decision.

For more details:

• Helen Toner discussed how Sam Altman provided the board with inaccurate information about OpenAI's safety processes on several occasions, undermining the board’s ability to oversee the company effectively.

• She cited specific incidents, including the board learning about ChatGPT’s release through Twitter and Altman’s undisclosed financial interests, which eroded trust.

• Toner mentioned Altman’s attempts to mislead board members about her, contributing further to the breakdown in trust.

• Ultimately, the cumulative effect of these actions led the board to conclude they could no longer trust Altman, making his position untenable.

Why this is important: This incident highlights the critical importance of transparency and trust in corporate governance, impacting OpenAI's future leadership and strategy.

NEW RESEARCH

💰 AI Outshines Analysts in Financial Forecasting

AI Researchs: A recent study reveals that GPT-4, a large language model (LLM), excels in financial statement analysis, surpassing human analysts in predicting future earnings trends.

For more details:

• Researchers at the University of Chicago provided GPT-4 with anonymized financial statements, asking it to analyze them for future earnings direction.

• GPT-4 outperformed human analysts in predicting earnings changes without needing narrative or industry-specific information.

• The model's prediction accuracy matched or slightly exceeded state-of-the-art machine learning models specialized in this task.

• The study showed that GPT-4’s insights stemmed from its economic reasoning and analysis of trends and ratios, not from its training data.

• Trading strategies based on GPT-4’s predictions resulted in higher Sharpe ratios and alphas compared to other models.

Why this is important: This study highlights the potential of AI to revolutionize the financial industry by providing superior analysis capabilities, democratizing access to expert-level insights, and transforming how market participants operate.

AI tutorial

✅ How to Access and Use Google multimodal Gemini 1.5 Pro.

AI Researchs: Google Gemini 1.5 Pro is now accessible in multiple countries, including the EU and the UK. This tutorial will guide you through accessing and utilizing the new Google Gemini 1.5 Pro model for various multimodal use cases.

Step-by-step:

1. Sign Up or Log In: Visit the Google AI Studio website and sign up or log in with your Google account.

2. Select Model: Once logged in, choose the Gemini 1.5 Pro model from the available options.

3. Create New Prompt: Click on "Create a new prompt," then choose between "Chat Prompt" and "Structured Prompt" based on your needs.

4. Configure System Instructions: For a chat prompt, enter specific system instructions (e.g., respond in a pirate-themed manner) and save the prompt.

5. Test and Save: Enter sample inputs and outputs to test the model's responses, then save the prompt for future use.

Pro tip: Use the structured prompt option for more consistent and tailored responses by providing detailed examples of desired inputs and outputs.

3 ChatGPT-4o prompts for job search

1. Cover letter writing:

Prompt: I'm applying for the [Job Title] position at [Company Name]. Can you help me draft a compelling cover letter that highlights my skills in [Your Skills] and my experience working remotely?

2. Optimize your resume

Prompt: I have a resume tailored for the [Industry]. Could you help me optimize it for remote job applications? Here is my resume [attach your resume]. What keywords should I emphasize, and how can I make my experience more relevant to remote work?

3. Finding remote jobs

Prompt: I have skills in [Your Skills] and I am looking for remote job opportunities. Can you help me find job openings that match my skill set and experience level?

AI Research Roundup

A recent study comparing large language models (LLMs) with 100,000 humans reveals that LLMs like GPT-4 can outperform humans in creativity, particularly in divergent association tasks.

Wikipedia remains the crowning achievement of Internet 1.0. It powered the rise of search engines (which depend on it) & generative AI (trained on its data). Through it all, Wikipedia remains. This new paper shows ChatGPT has yet to impact Wikipedia use.

Recent trends show that AI could replace CEOs, as nearly 47% of surveyed executives believe in automating most or all of a CEO's duties.

A TCS study reveals that 86% of senior executives use AI to enhance revenue, with 69% prioritizing innovation. However, only 4% have fully transformed their business with AI, and many feel unprepared for widespread AI use.

The AI chart

47% of Americans say they have never even heard of ChatGPT

AI pointers

Elon Musk recently told investors that his AI startup xAI is planning to build a supercomputer to power the next version of its AI chatbot Grok, with nearly 100,000 Nvidia, H100 chips.

George Lucas says the use of AI in cinema is inevitable and trying to stop it is like wanting to stick with horses over cars.

Peter Thiel, the tech billionaire says AI is bad news for people for math skills and society is going to shift in favor of people with strong verbal skills.

• 🌟 CallFluent AI : A tool to create voice agents for automating business calls.

• ⚕️ Markero : A platform to automate digital marketing tasks with content creation and lead management.

• 🔍 Canyon : A tool to search jobs with resume building, application autofill, and interview prep.

• 📑 Intellisay : A tool to convert spoken tasks into an organized daily schedule.

• 📜 Novelcrafter : A tool to write, plan, draft, and organize novels.

• ✅ Relume : A tool for website creation with sitemaps, wireframes, and designs.

• ✳️ Retrovision App : A tool to create depth-maps from images

In recent OpenAI post: All ChatGPT Free users can now use browse, vision, data analysis, file uploads, and GPTs.

Elon Musk’s xAI has raised $6 billion to compete with ChatGPT, with a current valuation of $18 billion and plans to develop its own AI supercomputer by late next year.

Nvidia's stock has officially reached a record high market cap of $2.8 trillion, positioning the company just 5% away from surpassing Apple as the world's second-largest company.

Lmsys revealed new performance results for Google’s latest Gemini 1.5 models. Pro/Advanced moved into second place, closely following GPT-4o, and Flash is approaching GPT-4 on the leaderboard.

Vox Media and OpenAI have partnered to use Vox Media's content to improve ChatGPT and create new AI products for audiences and advertisers.

Google’s AI Overviews is frequently producing incorrect information due to “hallucinations”, a problem Google CEO Sundar Pichai attributes to inherent flaws in large language models, which currently have no solution.

Brainly, an AI learning platform, has introduced Test Prep, a new feature that aims to assist students in preparing for exams by creating tailored study plans and practice questions.

Samsung is integrating Galaxy AI into the upcoming Galaxy Watch lineup, enhancing health features for a more personalized and intelligent experience.