SurvAIval Insights #29_24

Apple and Nvidia stealing data, Strawberry and Qstar, and OpenAI malpractices

Author

Rafa TV
July 20, 2024

Bon jour, my ro-bro fellas!

Fast weekly summary

  • Apple and Nvidia stealing data from YouTube?

  • OpenAI working on projects about strawberries and stars ;)

  • Should we worry about the malpractices and real intentions of OpenAI?

My mini survAIval insight of the week

I am currently reading The 48 Laws of Power, by Robert Greene, and the laws “Get others do the work for you, but always take the credit” caught my attention and made me think about the problem with AI and, specifically, its relation with the arts.

It is a kind of chilling law, with 2 main ideas:

  • You must secure the credit of your ideas and creations for yourself and keep others from stealing it away, and

  • don’t do all the work by yourself, let others do it for you, and this means, use others ideas for your own good.

It seems like the perfect law for the times we are living in.

In the coming years, there’s going to be an explosion of creativity in all industries, in all levels imaginable. This explosion is going to be based on the use of other people’s material (how AI companies train their models, potentially infringing copyright laws) and how available to everyone this new technology is.

Facing this unstoppable tech wave, do we have any other option?

How are we going to secure the credit of our creations?

In the middle of the vast amount of new content that is going to be created, it is going to be interesting to see how Robert Greene’s controversial law could be a historically proven useful application, or it will have to be reimagined from the ground up.

The chart of the week

As you can see, OpenAI seams to be the only AI company in the industry because they appear all the time in our weekly AI news, repeatedly. But, are they alone in this race?

The current ongoing battle for the “first place” in LLM capabilities is fought between OpenAI’s GPT-4o and Anthropic’s Claude 3.5 Sonnet. But the truth is that they should be seen as different use cases tools, in my opinion.

GPT-4o is better at quantitative capabilities. It is focused on being an “Agent” that will do stuff for us with the capacity to browse the web, create GPTs, create images, etc. Claude, on the other hand, is focused on an empathic human-like advisor we can relate to (even though we cannot talk to it, yet) precisely following complex and longer prompts, and an expert coder.

Another way of seeing this (maybe for a normal “real-world” users) is how “humanly” do these models sound in terms of language and tone, and writing capabilities.

As we can see in this chart, the winner is Claude.

If you want to have a chat with the most humanlike chatbot in the market right now, click in this link here and check it out by yourself!

My News Picks

My Video Picks

  • Another interesting use case of AI in the Internet world.

  • Dangers of AGI

  • New Sora demo from OpenAI and Tammy Lovin.

Tools and Education

  • A former OpenAI co-founder and ex-Tesla AI Director has launched a new type of AI school

  • Zapier believes computers should do more work for humans. So they distilled their AI products into a new digital energy drink: ZAP, AI in a can.

  • OneNote (Microsoft Copilot) can read and analyze your handwritten notes and use AI to improve them.

  • Fluent - Learn a new language by speaking with AI.

I hope these insights, news, and tools help you prepare for the future!

Have a really nice weekend and week.

Stay kind.

Rafa TV

🚨 Your feedback is valuable 🚨 Respond to this email and tell me how you think we could add more value to this newsletter. I really appreciate it!