- AI Fire
- Posts
- π€ Musk: No Tesla Share, No AI Access
π€ Musk: No Tesla Share, No AI Access
OpenAI dissolves AI safety team
Plus: OpenAI dissolves the AI safety team
Read time: 5 minutes
Great new day!
OpenAI's Superalignment team, led by Ilya Sutskever and Jan Leike, disbanded after their resignation. Core members are leaving. What's next for OpenAI in AI? Plus, ByteDance's Doubao is outpacing ChatGPT's 6.7 million US users. Find out more with us!
What are on FIRE π₯
IN PARTNERSHIP WITH VULTR
Tap into innovation with Vultr and receive instant access to cutting-edge NVIDIA GPUs for AI, VR, HPC, and More.
Vultr delivers immediate access to the latest NVIDIA GPUs for AI/ML, AR/VR, HPC, VDI/CAD, and more, available as on-demand virtual machines or large-scale dedicated clusters. Deploy worldwide without the complexity or cost of hyperscale providers. Plus, spin up servers in seconds for any purpose with Vultr Cloud Compute. Visit Vultr today and receive $250 in credits.
AI INSIGHTS
π Planning for AI Skills Now
Source: IDC
Enterprises must develop a holistic skillset, integrating AI capabilities across all roles.
Key Takeaways:
40% of enterprise respondents believe GenAI investments will continue.
62% of IT leaders reported missed revenue growth due to skill shortages.
$5.5T in losses predicted by 2026 due to skill shortages.
90% of organizations will face skill shortages by 2026.
AI tools are used to create and update training courses.
Why it matters: Investing in AI-driven training is essential to prevent financial losses and improve workforce readiness in a rapidly evolving tech landscape.
PRESENTED BY MAX AI
MaxAI.me - Outsmart Most People with 1-Click AI
Discover MaxAI.me, one of the top 50 GenAI apps of 2024!
Best features:
Chat with the latest AI like GPT-4, Claude 3, and Gemini 1.5, all in one place.
Perfect your writing anywhere with 1-click AI without copy-pasting.
Save 90% of your reading & watching time with AI summaries.
Reply 10x faster with AI on email, social media, and messaging web apps.
Rapidly turn your visions into stunning images with AI art generators.
AI MODEL
π‘ New Hard Prompts in Chatbot Arena
Source: LMSYS
Hard Prompts in Chatbot Arena provide a new way to test advanced language models.
Key Takeaways:
The new category features user-submitted challenging prompts.
Prompts are evaluated based on domain knowledge, complexity, and problem-solving.
Llama-3-8B-Instruct drops in ranking compared to GPT-4-0314 in this category.
Claude-3-Opus surpasses Llama-3-70B-Instruct in the new ranking.
GPT-3.5-Turbo-1106/0125 and Claude-2.1 show notable improvements.
Why it matters: Hard Prompts help advance AI research by exposing models to more complex scenarios, driving improvements in their reasoning and problem-solving capabilities, which are crucial for future AI developments.
TODAY IN AI
AI HIGHLIGHTS
π§ Study shows GPT-4's human-like performance in theory of mind tasks, including false belief and strange stories. Researchers tested humans and LLMs, noting AI's limitations and the need for cautious interpretation.
π» Microsoft's new Copilot+ PCs feature AI capabilities, including a new Recall feature and enhanced AI apps. Surface laptops with Snapdragon X Elite chipset boast up to 22-hour battery life. Third-party brands like HP and Dell also join the AI PC wave.
Real-time conversations with your AI companion Copilot, powered by OpenAIβs GPT-4o.
Coming Soon: Copilot π€ Minecraft
β Microsoft Copilot (@MSFTCopilot)
6:38 PM β’ May 20, 2024
ποΈ OpenAI removed the 'Sky' voice after Scarlett Johansson objected to its similarity to her voice. Johansson had declined to participate in the project. OpenAI stated that the voice belongs to a different professional actress.
π€ Doubao by ByteDance surpasses Baidu's Ernie Bot in downloads and active users. With 26 million monthly active users, Doubao highlights ByteDance's AI advancements, outpacing ChatGPT's 6.7 million users in the US.
π OpenAI's Superalignment team, focused on AI risks, was disbanded after leaders Ilya Sutskever and Jan Leike resigned. Leike cited concerns over safety being sidelined for product development. Sam Altman expressed regret over their departure.
i'm super appreciative of @janleike's contributions to openai's alignment research and safety culture, and very sad to see him leave. he's right we have a lot more to do; we are committed to doing it. i'll have a longer post in the next couple of days.
π§‘
β Sam Altman (@sama)
6:56 PM β’ May 17, 2024
Leike mentioned that he and the company had reached a βbreaking pointβ because he believed they weren't investing enough in preparedness.
βThese problems are quite hard to get right,β Leike wrote on X. βAnd I am concerned we arenβt on a trajectory to get there.β
According to sources cited by TechCrunch, the super alignment team was severely under-resourced β despite OpenAIβs promise to dedicate a fifth of its compute to safety.
Do you think Leike and Sutskever will be the last members to leave OpenAI, or will more follow? |
π Elon Musk threatens to withhold AI and robotics products unless he gets 25% control of Tesla. Musk, currently holding 13%, insists these conditions are vital for Tesla's success, including reincorporating in Texas.
AI STOCK
DAILY AI FUNDRAISING
Overland AI, a Seattle-based company, raised $10M in seed funding led by Point72 Ventures. The funds will enhance their OverDrive software for autonomous off-road vehicles, supporting U.S. Army projects.
AI SOURCES FROM AI FIRE
NEW EMPOWERED AI TOOLS
β TestSprite delivers AI-powered, fully automated end-to-end testing solutions.
πΌ Buffup.AI is an innovative and extremely user-friendly AI assistant with ChatGPT-4o
π ChatPlayground AI achieves better AI answers 73% of the time with multi chatbots
π Phew AI Tab manages and retrieves tab information via AI-based grouping and spaces in a vertical sidebar.
βοΈ fforward AI analyzes customer interviews and discovers valuable opportunities that tie your customers together.
AI QUICK HITS
1. π Anthropic Enhances AI Policy, Gains Key Insights (Read more)
2. π Gemini 1.5 Boosts Understanding Across Many Tokens (Read more)
3. π‘ Could Light Chips Boost AI's Computing Power? (Read more)
4. π The School is βTrying To Embrace AIβ As Much As Possible (Read more)
5. β οΈ DHS Alerts AI Risks to Election Security (Read more)
AI CHARTS
If you haven't reviewed your LLM choices and uses in the last 6 months, it's time to check again. In October, we paid $30 input and $60 output per million tokens for GPT-4 responses. Now, there's a cheaper GPT-4o model at $5 input and $15 output per million tokens, plus a 50% discount for batch processing.
Don't overlook the batch option; it usually provides responses within a few hours, though it can take up to 24 hours.
Action Steps Now?
Review Current Use Cases: Check if your current applications are cost-effective and performing well.
Explore New Opportunities: With lower costs, think about using LLM for more tasks.
Leverage Batch Options: Use the 50% discount for big projects.
Make sure your LLM choices give you the best value and performance.
AI TUTORIAL
Find & Replace The Object On Getimg
Visit Getimg.
Select the 'Tool' section and choose 'Image Editor.'
Use 'Find & Replace,' describe what you want to change, and click the button to generate.
AI PROMPT
A Prompt to Communicate Effectively with Narcissists at Work
Generate a detailed list of dos and don'ts for effectively communicating with a narcissist at work, specifically tailored to my business. My business, [describe your business], operates in the [industry/sector] and focuses on [key activities or services]. We have a team of [number of team members] who are [describe team dynamics, roles, and responsibilities], including [mention any specific roles or hierarchy].
The primary goal is to ensure clear communication and minimize misunderstandings, thus maintaining a smooth workflow and a respectful work environment. The guidance should offer practical steps for engaging with a narcissist without disrupting team dynamics, aiming to enhance overall productivity and workplace harmony. Consider including strategies for:
Setting and maintaining boundaries.
Providing feedback constructively.
Managing conflicts effectively.
Encouraging teamwork and collaboration.
Ensuring that all team members feel valued and respected.
Please ensure the advice is actionable and relevant to the context provided, helping us navigate interactions with a narcissistic colleague while preserving the integrity and efficiency of our operations
AI-GENERATED IMAGES
Midjourney Sref Code 480 for Hand-Drawn Illustration
prompt: [your object] --sref 480
AI JOBS
SPONSOR US
Bring your product/service to the AI minds of 300K+ audiences across our platforms in top-tier countries
We offer a wide range of sponsorship opportunities to match your needs and budget, maximizing your brand's presence and connecting with our audience. Make magic happen here.
We read your emails, comments, and poll replies daily
How would you rate todayβs newsletter?Your feedback helps us create the best newsletter possible |
Reviews of the day
LAST WORD π
How are we doing?
We love hearing from readers and are always looking for feedback. How are we doing with AI Fire? Is there anything you'd like to see more of or less of? Which aspects of the newsletter (or AI Fire Academy) do you enjoy most?
Hit reply and say Hello β we'd love to hear from you!
Like what you're reading? Forward it to friends, and they can sign up here.
Cheers,
The AI Fire Team
Reply