Welcome to AI Tool Report!

๐Ÿ˜ฎ ChatGPT getting lazier? OpenAI confirms

๐Ÿ‡จ๐Ÿ‡ณ Chinaโ€™s WeRide tests autonomous buses in Singapore
๐Ÿง  ChatGPT passes neurology exam for first time
๐Ÿ‘™ Sports Illustrated fires CEO after AI drama

_______________________________________________________________

Read Time: 4 minutes

Our Report: OpenAI has confirmed that ChatGPT is getting 'lazier' after many users reported a severe dip in the quality of the responses they were receiving from the model. Will users adopt Claude, Gemini, or some other AI chatbot in the meantime?

๐Ÿ”‘ Key Points:

  • OpenAI has acknowledged a dip in GPT-4's performance, with users reporting decreased efficiency in task completion stemming from a lack of updates since November 11th.

  • Notably, the performance degradation only affects specific prompts in ChatGPT (though these prompts have not been disclosed), and whilst investigations continue, sources are unclear whether performance will be impacted in models like Microsoftโ€™s CoPilot & Bing.

  • OpenAI is supposedly actively working on a solution, but the timeline for the fix remains uncertain. While GPT-4 undergoes maintenance, users may adopt similar services in Gemini, Claude, Mistral, Bard, or any other AI chatbot out there.

๐Ÿคจ Why you should care: As GPT-4 is a key player in the AI industry, its performance issues could affect a wide range of applications and users (potentially all Microsoft & Bing users)โ€”potentially reshaping the AI chatbot landscape and distributing the user base amongst more AI models.

In Partnership with OctoML

OctoAI just released blazing-fast, ready-to-use endpoints in their new Text Gen Solution, allowing you to scale seamlessly as your LLM app takes off. Maybe you started building with GPT? Great. But itโ€™s not always the best, fastest, or cheapest option for everything in your app. If youโ€™re early on, you need the flexibility to explore open-source LLMs and reduce your dependency on a single closed-source provider.

With OctoAI:

  • Build on your choice of Llama 2-Chat, Code Llama Instruct, or Mistral Instruct models

  • Bring your fine-tuned model to run on their super-performant infra

  • Leverage a โ€œmodel cocktail,โ€ running open source alongside OpenAI

  • Build with an API compatible with OpenAI

Sign up for OctoAI today and instantly get $10 for your next project, or contact their team to discuss opportunities for fully funded proof of concept (POC) projects.

Read the blog to get started

  1. Seamless helps you draft literature reviews 100x faster with AI

  2. Shotty transforms URLs into Instagram-like shorts

  3. Mistral is a base model ready for fine-tuning

  4. Radiant is ChatGPT for healthcare learners

  5. Kards generates instant flashcards with AI

โœ… Web Design Consultant

copy & paste โฌ‡๏ธ

โ€œI want you to act as a web design consultant. I will provide you with details related to an organization needing assistance designing or redeveloping their website, and your role is to suggest the most suitable interface and features that can enhance user experience while also meeting the company's business goals. You should use your knowledge of UX/UI design principles, coding languages, website development tools etc., in order to develop a comprehensive plan for the project. My first request is "I need help creating an e-commerce site for selling XYZโ€

๐ŸŽจ Artist Highlight: Ralph Lentjes

GPT-4.0 has successfully passed a neurology exam from the American Board of Psychiatry and Neurology, answering 85% of the questions correctlyโ€”suggesting significant potential applications in clinical neurology.

The test, involving questions also from the European Board for Neurology, highlighted ChatGPT-4.0's strengths in behavioral, cognitive, and psychological areas; however, the model showed limitations in higher-order thinking tasks. Collectively the results signify a renewed optimism for potential medical applications, although it may be a while before we see it actively used in a practical setting.

The Arena Group (owner of Sports Illustrated) has dismissed CEO Ross Levinsohn following an AI-related scandal, where Sports Illustrated published multiple AI-generated articles under false author names and photos.

This was part of a larger corporate shakeup as the group also terminated three senior executives, aiming to restructure and improve operations, whilst details linking the AI issue to these dismissals remain unelaborated by the company (seems like an excuse to fire particular people more than anything).

Chinese autonomous vehicle company WeRide is expanding globally, recently obtaining licenses in Singapore for large-scale public road testing of its autonomous buses, which enable testing in key areas like the One North tech cluster (Singaporeโ€™s Silicon Valley).

This move follows WeRide's license acquisition for robotaxi road tests in the UAE and reflects a strategic shift from Level 4 robotaxis (human oversight) to more manageable autonomous buses.

Remote, SF / $Competitive / Have to be located <50 miles of SF

โ€œAs an Enterprise sales manager, you'll be responsible for building and leading a high-performing enterprise sales team that targets CIOs and CMOs at Fortune 500/Global 2000 companies, providing guidance, coaching, and support to drive individual and team success. Your primary focus will be on driving revenue growth from both new logos and existing accounts, establishing and iterating on repeatable sales processes, and helping Writer continually improve how we serve our enterprise customers.โ€

Remote / $ Competitive / Choose your own hours

โ€œWe are a growing start-up company looking for an AI Virtual Assistant to join our community. As an AI Virtual Assistant, you will be responsible for a variety of tasks to help our community, including researching, sharing, and mentoring other VAs to succeed with using Gen AI and AI Tools. Our community encompasses content creators, freelancers, independent workers, gig workers, resellers & thrifters, entrepreneurs, solopreneurs, and small business owners.โ€

CA / $80ph / Choose your own hours

โ€œWe offer pay-per-task projects that involve (not limited to) creating/collecting short sentences or texts, capturing images and videos, or featuring short video captures of participants' faces or movements. For some projects, you might be asked to come on-premises, but some tasks can be completed remotely from homeโ€

The AI Tool Report just became the fastest-growing AI newsletter in the world, with 450,000+ readers working at companies like Apple, Meta, Google, Microsoft, and many more. Weโ€™re now booked out 4 weeks in advance, due to a massive surge in demand. Book your ad spot before someone else doesโ€ฆ

๐ŸŽจ Pika 1.0 officially available to all users

๐Ÿ‡ซ๐Ÿ‡ท MistralAI closes its $415M funding round

๐Ÿ‡ช๐Ÿ‡บ EU says general-purpose AI rules can evolve over time

๐Ÿคซ Ashleyโ€”the worldโ€™s first AI-powered political campaign caller

๐Ÿ˜ฎ Investingdoctcom caught plagiarizing financial news with AI

We read your emails, comments, and poll replies daily.

Hit reply and let us know what you want more of!

Until next time, Martin & Arturo.

Keep Reading