MindStudio Trainer on its way + Workshop Recording

Once again, this week led to new heights for both our platform and the AI space as a whole.

French startup Mistral reached a $2b valuation and released Mistral Large, Gemini faced backlash for being too censored, and MindStudio released the biggest pack of UI changes to date.

Now, it’s all about what YOU will do with your newfound AI powers. Keep reading to learn more and find some suggestions to stimulate your mind.

Congratulations to our community members and creators that contributed to making this week special!

🗞️ Industry news

1. Mistral launches Mistral Large & Proprietary Chatbot
   Mistral, the $2b AI French unicorn, released Mistral large this week. Contrary to Mistral small and Mistral medium, the large model isn’t currently available under an open source license.
   

   

   
   According to the company and common tests, Mistral Large performs nearly as well as GPT-4. The model is now available in Microsoft Azure, and you can test it out in their new chatbot “Le Chat”.
   

2. Sundar Pichai weighs in on criticism of Gemini being too censored
   In one of our training videos, we talk about which model is your likely best bet depending on use case. In the video, I mention Gemini being overly censored, hence unable to provide useful answers across many domains.
   
   Many users started to notice the same, and the CEO had to weigh in when Gemini Pro started generating images of inaccurate historical events or cultural figures. The team agreed the model may overcompensate for diversity and inclusion and shut off the image generation capabilities in the chatbot.
   
   Google has always played catch up since OpenAI released ChatGPT back in November 2022. It’s crazy to think they were actually the company releasing the research on transformers, the very same technology that made the AI revolution in 2023 possible.
   
   Sundar Pichai agreed this is unacceptable and that Google should focus on organizing the world’s information.
   

3. Is coding dead? Analysis by Matt Wolfe
   

   

   
   Matt Wolfe shared this graph from Tecky Academy, together with snippets of NVIDIA and StabilityAI’s CEOs, in his latest video on AI + coding.
   
   Up until 2y ago, learning to code was considered one of the best possible choices for any professional career, or at least careers that at some point interact with a machine.
   
   With AI, given the average kid (e.g. 5y old) will enter the workforce in 10y+ from now, that’s not such a safe bet anymore. MindStudio’s CEO Dmitry Shapiro also considers the ability to communicate with AI models now nearly as important as coding itself.

🔥 Product Updates

Last week, we introduced several exciting features to further improve the platform:

• Uniform AI: you will notice a huge change in how every element in the platform feels and operate. We deployed a coherent design (100s of small changes) that should make a notable difference in your daily workflows;

• New “Experimental” Logic expression: previously, the logic block only worked with GPT-3.5. The new experimental engine uses GPT-4 turbo, better suited for complex evaluation tasks. I used it for our semantic menu in MindStudio trainer;

• Fallback models: we now report on the availability and status of all models in our platform. If any model fails to retrieve data, or is currently unavailable, MindStudio will automatically fall back to a reliable alternative during the breakout;

• Updated GPT-4 Turbo: MindStudio updated GPT-4 Turbo from the old preview to the new -0215 endpoint.

• Workshop recording: you can find the recording of the latest Zapier Workshop here.

As we move forward, our team is dedicated to delivering even more enhancements and functionalities. Next week, we’ll focus on:

• Announcing the new packages: we know many of our advanced builders are waiting to see what the new structure looks like to price & charge for access to MindStudio applications. We’re working hard to announce the change as soon as possible next week;

• A LOT more training resources: we’re working on a 6-week workshop with 2 classes each week to get from 0 to hero. The first cohort will experience it live, while the next ones will have a mix of recorded and live sessions. We’re committed to ensure anyone wanting to build AI applications can do so in MindStudio - you’re not alone!

• Releasing MindStudio trainer: while the first draft is ready for testing, our MindStudio trainer is still very much in Alpha and can output incorrect or inappropriate answers. I’ll work on it to ensure stability and reliability. Once ready, we’ll release it to the broader community.

💡 Tip of The Week

You’re probably underestimating the power of Mistral. This week, I tested all the models in our platform for speed and matched the rankings with HuggingFace’s Arena ELO.

These are the results:

GPT-4 and GPT-4 Turbo dominated the rankings, but are pretty slow. At 69.238 tokens per second, Mistral Medium is the over 2x times faster than GPT-3.5 AND performs nearly as well as GPT-4 turbo according to 1000s of ratings on HuggingFace.

Mistral Medium is up to 10x cheaper than the top ranking models, and can be a great assist in saving costs and/or spend the same with larger prompts.

🤝 Community Events

If you’re looking to meet our Product Manager and group up with community members, you can join our Discord Hangout on Friday 1st @ 3pm EST. Join from this link.

🌯 That’s a wrap!

AI is moving faster than usual, and we’re happy to be part of this groundbreaking revolution. Stay tuned to learn more about what’s next and get tips & tricks for your MindStudio build.

You saw it here first,

Giorgio Barilla
MindStudio Developer & Project Manager @ MindStudio