🧑‍⚖️ OpenAI Cited Again for Data Breach

Hello AI Enthusiast,

Both OpenAI and Google are making big moves this week, forming key partnerships. OpenAI is also enhancing its models but faces legal checks in Italy, while Google showcases Lumiere, a new AI video tool, and joins forces with Hugging Face for AI development.

Before we dive into the news section, let's talk about an opportunity for those who subscribe to the waiting list for our fifth edition of the Master in Prompt Engineering, starting March 4th. We're hosting a webinar titled "How AI Academy Uses Artificial Intelligence and How You Can Too." It’ll take place on Friday, February 2nd, from 6 to 7 pm CET.

In this one-hour event, Gianluca, our CEO, will reveal how AI integrations contributed to our 300% growth last year. He'll demonstrate how we built and use these tools, and how you can do that too.

Given the rapid sell-out of our previous editions, we'll be offering early access to the new edition seats to our waitlist subscribers right after the webinar, before opening them to the general public.

Now, in our last newsletter, we asked for your take on a topic involving AI and creativity. To see what our community thinks scroll down for the poll results.

From our community 🤝

Last week we asked you for your thoughts on AI as a creative companion. These are the results:

Can AI assist the creative process?

🟩🟩🟩🟩🟩🟩 Yes, AI can enhance creativity. 🤖 (62%)

🟨🟨🟨⬜️⬜️⬜️ No, keep creativity a solely human endeavor. 🧑 (38%)

It looks like many of you think AI can be a good assistant, even for creative work, which we used to think only humans could do.

We also want to highlight some of the comments we got, which are pretty cool.

“If it is used ethically and collaboratively, it can enhance our own creativity, and help to open up new approaches. There is a legalitimacy to the copyright issue, as its output is based upon the "art" and "creative production" of others. But then again, we, as humans, do the same thing. We are all influenced by what has come before us. ”

Answer this week’s poll and see how your opinion compares with the community next week!

News Bytes 🗞️

• OpenAI and Common Sense Media (a US’s leading advocacy group for children and families) have partnered to develop AI guidelines and educational resources for families and schools. OpenAI is trying to set the standards for how their technology gets used, often anticipating governments.

  What do you think about OpenAI's educational partnership?

  Good: It ensures safe and ethical AI use in education and family settings. Concerning: It might influence the use of AI in educational spaces.

  Login or Subscribe to participate in polls.

• OpenAI goes “better, faster, cheaper”: it’s released new more accurate embedding models¹ and has reduced the prices of its GPT-3.5 Turbo model, used widely in text-intensive applications, by 50% for input and 25% for output. Additionally, they have released a new model version for the chatbot ChatGPT and are working on launching GPT-4 Turbo soon. You can find more info about this in our “Educational Pill“ section below.

• Italy is angry at OpenAI (again) and is investigating them for GDPR violations. The specific details of the breaches have not been disclosed, but OpenAI may face fines and orders to change its data processing practices. OpenAI has 30 days to respond, we’ll keep you updated.

• Google has showcased Lumiere, an AI video generation model that creates animated videos from a text prompt or image. The model is still under research, so far we have just a scientific paper and some nice demos. We’re still skeptical as there’s always a huge difference between research and products (especially after the Gemini demo…), but in the meantime check out this demo of “A dog driving a car on a suburban street wearing funny sunglasses”.

  

• Hugging Face has announced a strategic partnership with Google Cloud to enhance AI development. The collaboration aims to make the latest AI research accessible through Hugging Face's open-source libraries and enable easy training and deployment of Hugging Face models within Google Cloud's infrastructure. This partnership will benefit developers, and organizations looking to build their own AI models.

• The team behind Google DeepMind has proposed an AI system called The Articulate Medical Intelligence Explorer(AMIE). Deepmind had already shown “better than doctors” performance in the lab with their AI model MedPalm 2, but AMIE is much closer to being used in the real world. The main differences are that AMIE is a conversational interface (much easier to use for a patient) and has been tested in a real clinical trial.

• The Federal Trade Commission (FTC) has sent letters to Alphabet, Amazon, Anthropic, Microsoft, and OpenAI, requesting information on their investments and partnerships. The FTC is worried about their relationships: Microsoft competes with OpenAI but has invested in them, and Google competes with Anthropic but gave them billions (with a B). Is it possible that they’re using their power as investors to dictate their competitor’s agenda?

• A study on Replika, an AI chatbot, revealed that 90% of the participants suffered from loneliness, with 43% qualifying as severely lonely. Surprisingly, 30 individuals who were found to suffer from depression credited Replika with preventing their suicidal actions. We’ve talked about the problem of users falling in love with these apps (and the power that give to the companies behind them), but we can’t overlook the possible positive impact for people who are suffering. This is a highly controversial topic and raises a big question. of whether AI friends are a good way to help with loneliness.

  Are AI companions effective for loneliness?

  Yes, they provide valuable support. No, they're not a substitute for human interaction.

  Login or Subscribe to participate in polls.

• The White House has called for legislation to tackle the issue of fake sexual images generated by AI, in response to the proliferation of such photos featuring Taylor Swift on social media that were viewed millions of times before removal. The government body emphasized the vital role social media platforms play in enforcing rules to prevent the spread of such content.

• OpenAI has suspended a developer who created a ChatGPT-powered bot, Dean. Bot, designed to impersonate Dean Phillips, a Democratic political candidate, and assist in his campaign. The bot's use, although it was clearly stated that it was only impersonating the politician, violates OpenAI's policies on applications for political campaigning and lobbying. The US election contest is firing up. 🔥

This week’s glossary 📖

1. Embedding Models: Algorithms used in machine learning and natural language processing to represent words or phrases as vectors of numbers, helping computers understand language.

Educational Pill 💊

OpenAI's Model and Pricing Updates

OpenAI has recently made significant updates to its models and pricing. They've notably reduced costs for their GPT-3.5 Turbo model, cutting input prices by 50% and output by 25%. This is a big deal for those using AI for text-heavy tasks, as it makes advanced AI tools more accessible and affordable. Additionally, OpenAI has enhanced the model to improve performance, ensuring it remains a top choice for developers and users.

Moreover, OpenAI introduced an updated version of GPT-4 Turbo, addressing cases of “laziness” where the model would occasionally not complete tasks. Alongside this, they have released improved text embedding models.

Gianluca has explained it in 5 easy to understand points here.

LOLgorithms 😂

Are we the only ones to find this creepy?

That is the end of our newsletter.

Remember, if your company is looking to implement AI technologies, we also offer customized corporate training.

See you next week! 👋