The Generative AI newsletter for legal pros everywhere.

👋 Hello! to our NEW SUBSCRIBERS! To read previous posts, go here.

🥌  To all: Thank you for reading, sharing, and engaging! You keep the pressure on me to deliver every day. I won’t let you down!

Ok, today we will:

1. review current regulatory and legal actions

2. show you an easy-to-create “manual auto AI agent”

3. share a video demo of a legal contracting tool

4. read about The Sims game but for Auto AI Agents

Legal Actions Update

Let’s take a look at current regulatory, executive, and matters being pursued around the globe regarding ChatGPT and GenAI.

• Italy is the only one to outright ban ChatGPT 

• US has an AI Bill of Rights being worked on plus the Commerce Department put out a request for public comments earlier today as the Biden administration seeks to shape specific policies - this follows on the heels of the USPTO’s own comment-seeking announcement

• EU works on The AI Act

• UK is working to regulate AI somehow 

• The Office of the Privacy Commissioner of Canada has launched an investigation into Open AI

• Ireland is looking into AI regulation with its Irish Data Protection Commission looking at Italy for ideas and will coordinate with the EU

• Germany said a ban similar to those in other nations could follow if OpenAI violates GDPR or similar policies

• China has unveiled new rules regarding AI

• Meanwhile, Japan is embracing ChatGPT while many Japanese universities regulate it

Some known legal actions:

• Center for AI & Digital Policy filed a complaint with Federal Trade Commission against OpenAI

• Apple hit with lawsuit over fake chatbot on its Appstore

• An Australian mayor may sue OpenAI for defamation over inaccurate information about him provided by ChatGPT

• GitHub is facing a class action lawsuit over the legal rights of the creators of the open-source coding in Copilot training data

• A related lawsuit targeting Stability AI, DeviantArt, and Mid-Journey face a class action lawsuit over the use of StableDiffusion, which used copyrighted art in its training data

• Getty Images filed against Stability AI for using Getty Images’ copyrighted content in its training model

Use Case: Manual Auto AI Agents

Recently I shared the news about Auto-GPT and AI Agents that are being developed to self-prompt, use multiple LLMs, and access various resources (internet, data, etc.) to accomplish a set of goals. They work by themselves until they reach milestones (meaning less human input is needed 😱).

There is also a news story below on these (blown mind 🤯)

Today I thought I would give you a small taste of this in the ChatGPT you have access to.

We are going to create a “manual auto AI agent” - if that makes sense - it will. 😉

Here is the first prompt.

▶︎▶︎PROMPT: For the following prompt, I want you to respond with the answer but then immediately generate a new prompt to refine or iterate on your answer to make it better. Here is the seed prompt: [What are the basics of the billable hour in the legal profession?]

[ ] = Whatever topic/question you choose

Notice what I am asking ChatGPT to do here. I want its answer but then I want it to review the original prompt and its answer in order to refine or iterate on it in order to create another prompt.

The way this plays out is like this. 👇

1. ChatGPT will provide its answer to the prompt and then create a new line that says “New Prompt . . .”

2. Now ChatGPT cannot prompt itself, so you have to (thus why this is manual auto). Copy-n-paste the first part of the original prompt: “For the following prompt, I want you to respond with the answer but then immediately generate a new prompt to refine or iterate on your answer to make it better. Here is the seed prompt:” Then paste in the new prompt that ChatGPT gave you.

3. Keep doing this and you will get a sense of how Auto-GPT works to some degree.

4. Also, this is a great way to have ChatGPT structure a learning journey for you as it is creating the follow-up questions for you.

Here is an example you can review.

News you can Use: 

5-year old legal tech company Rally has made the product Spellbook. One of the co-founders is a lawyer. Here is a nice video overview with a demo of the product. (This is not an ad and I get no $ for this just so you know.)

More News you can Use: 

A new research paper discusses how the authors introduced computer programs called generative agents that imitate realistic human actions. These agents can perform everyday tasks, have hobbies, form opinions, interact with each other, and even plan their days. They created a system that lets these agents store and use their experiences to decide what to do next.

The authors put the agents in a virtual world like The Sims, where users can chat with them using everyday language. In a test, the agents showed lifelike behavior by planning a party, inviting others, making friends, and going on dates.

They found that observing, planning, and reflecting are all important parts of making the agents' behavior more believable. This work combines advanced language models with interactive agents to create more realistic simulations of human behavior.

Even More News you can Use: 

That's a wrap for today. Stay thirsty & see ya next time! If you want more, be sure to follow me on Twitter and LinkedIn.

DISCLAIMER: None of this is legal advice. This newsletter is strictly educational and is not legal advice or a solicitation to buy or sell any assets or to make any legal decisions. Please be careful and do your own research.