Let's rewind to the heyday of Napster, when a whole new universe of music unfolded before our very eyes, sparking an insatiable hunger for sonic exploration.

As we journey into today's newsletter, we'll venture into the fascinating frontier where copyright clashes with the innovative realm of generative AI.

Buckle up and join us as we navigate the intricate maze of creativity, law, and technology that's shaping the future of the digital landscape! 🚀

A special Hello 👋 to my NEW SUBSCRIBERS! 
To read previous posts, go here.

Ok, today we will:

1. update you on music industry napster-slayers

2. talk more copyright legal actions

3. discuss what websites are in an AI training model

4. say that copyrights kill creativity

5. show you the tuning dials for ChatGPT responses

6. share how legal tech is goin’ AI (and getting $$)

7. suggest you don’t bet your future on becoming a prompt engineer

What’s Yours Is Mine Is Yours (Again)

Fair-use. Human-generated. Hybrid-generated. We are in the midst of another mind-bending journey into who (or what) creates content, what rights they (or it) have, and who and how it might be used by others. In other words, here we go again.

As AI-generated art and text gain traction, courts and regulators are grappling with intellectual property law to address two main issues:

1. the use of copyrighted data to train AI models, and

2. the potential copyrighting of AI-generated products.

AI-Music Battle: UMG's Napster Encore

In a remix that's part déjà vu, part high-stakes symphony, the music industry is dusting off its battle gear and gearing up for a war against AI-generated music. Universal Music Group, the powerhouse behind an impressive one-third of the global music market, is strumming up a storm as it urges streaming platforms like Spotify to hit the mute button on AI-driven copycats.

Channeling their inner Napster-slayers, UMG is requesting Apple Music and Spotify to pull the plug on AI developers while orchestrating takedown requests for YouTube users who dare to mimic their artists. Just as the industry once vanquished peer-to-peer sharing sites like Napster, it's now eager to see whether AI developers can continue sampling copyrighted works without permission.

As the overture begins for this impending AI showdown, I can't help but ponder if the music industry might someday strike a chord with AI developers, harmonizing commercial opportunities with creative, AI-generated tunes that have us all singing along.

We will get into the creativity vs copyright debate below, but first, let’s tackle some related topics.

Current legal battleground

The legal battleground has seen cases such as the lawsuit against AI art generators Stable Diffusion and Midjourney, filed by artists claiming infringement of their rights. Artists Sarah Andersen, Kelly McKernan, and Karla Ortiz sued the companies in January, claiming that the unauthorized copying of their works to train the systems and the creation of AI-generated images in their styles violated their rights. Getty Images has also sued Stable Diffusion's developer for using their images without consent.

At the heart of these cases is the key legal principle of "fair use," which permits limited use of copyrighted material for specific purposes. Companies like Stability AI and OpenAI argue that their use of images falls under fair use as it is transformative.

In March, the U.S. Copyright Office released new guidelines regarding copyrighting AI-generated images, stating that eligibility depends on the circumstances, the AI tool's operation, and its use in creating the final work.

Current intellectual property law requires human authorship for patents, trademarks, and copyrights, complicating the issue of copyrighting AI-generated works.

In response, earlier this week, to the group of artists alleging widespread copyright infringement through the exploitation of their work in generative AI systems, Stability AI, Midjourney, and DeviantArt have requested that a federal court in San Francisco reject the artists' proposed class action lawsuit. The companies argue that the AI-produced images are not comparable to the artists' creations and that the case lacks precise details about the photos allegedly misappropriated.

Training Data

The Washington Post released its finding from an investigation into the dataset used to train AI chatbots, such as Google's C4 dataset. It revealed an assortment of websites that contribute to AI training data.

Key findings include the presence of potentially biased, unreliable, and copyrighted content, raising concerns over AI's potential for spreading misinformation or violating intellectual property rights.

Other Findings:

• Google's C4 dataset, a massive snapshot of 15 million websites, is used to instruct some prominent English-language AIs, including Google's T5 and Facebook's LLaMA.

• The C4 dataset includes websites from industries such as journalism, entertainment, software development, medicine, and content creation.

• The top three websites in the dataset are patents.google.com, wikipedia.org, and scribd.com.

• Privacy concerns arise with the inclusion of websites like coloradovoters.info and flvoters.com, which contain state voter registration databases.

Related to this, Elon Musk created a stir on Twitter (shocking right?) when he replied to Microsoft dropping Twitter as an advertising channel. He is going to sue them for using Twitter in one of their AI training model data sets.

Copyright The Creativity Killer: A Contrarian POV

Ah, copyright! The venerable shield that protects the livelihoods of countless artists and creators, guarding their works against the marauding hordes of plagiarists and intellectual property pirates.

But, what if I told you that the very shield meant to protect creative expression might, in fact, be stifling it? As the use of generative AI tools proliferates, it's time to take a closer look at copyright.

The technological revolution has given birth to countless innovations and new creative avenues, but it has also sparked endless debates around the future of creativity and copyright. Most arguments revolve around the supposed battle between artists and technology developers. But, in reality, this narrow frame of reference fails to consider the full spectrum of stakeholders involved, particularly those who both create and utilize these AI tools.

Historically, the advent of new technologies has often been met with fear and skepticism. From the camcorder to hip-hop, tools that initially seemed to threaten art and artists have eventually fostered innovation and expanded the creative landscape. Today's generative AI technology is no different. Far from being mere machines that churn out subpar content, AI tools have been aiding professionals and amateurs alike in developing their artistic expression, from providing insights into writer's block to assisting in the creation of commercial artwork.

The true challenge lies in navigating the murky waters of copyright policy when it comes to generative AI. The fact remains that every creator is influenced by the works that came before them, regardless of the medium. It's crucial that we approach copyright policy with this in mind, permitting AI tools that learn from past works in ways that facilitate the creation of new, distinct ones, without infringing upon or substituting for preexisting expressions.

It’s time to take a step back and reassess our relationship with copyright. While it may be tempting to view it as the ultimate protector of artistic endeavor, it's essential that we recognize its potential to over-protect and curb creativity. By embracing the potential of generative AI and striking the right balance in copyright policy, we can foster a new era of creative expression that empowers both existing and future creators.

Use Case: Tuning ChatGPT’s Dials

When working with advanced AI models like ChatGPT, you can control the model's output behavior by adjusting a parameter called "temperature."

In plain language, the temperature setting helps you manage how creative or focused the AI's responses will be. In this explainer, I will show you what temperature is, how to use it, and provide some examples to help you understand its effects.

What is temperature?

Temperature, in the context of AI models, refers to a parameter that influences the randomness or creativity of the generated text. A high temperature encourages the model to explore more diverse and creative possibilities, while a low temperature guides it towards more focused and deterministic responses.

Think of temperature as a dial that helps you control the balance between creativity and consistency in AI-generated text.

How to use temperature:

To use temperature, you simply adjust the temperature value when providing a prompt to the AI. The value ranges from 0 to 1, with 0 being the most focused and deterministic, and 1 being the most creative and random. Typically, a value of around 0.7 provides a good balance between creativity and coherence.

Examples:

Let's see how temperature affects the AI's responses to the same prompt. We'll use the prompt, "What could be a possible future invention?" with different temperature settings:

Temperature: 0.2 (Low)

AI response: "A possible future invention could be self-driving electric cars that communicate with each other to optimize traffic flow and reduce accidents."

Temperature: 0.7 (Medium)

AI response: "A possible future invention might involve a smart fabric that can change its properties based on the wearer's needs, like adapting to weather conditions or even healing wounds."

Temperature: 1.0 (High)

AI response: "In the future, we might invent a thought-to-taste synthesizer that allows people to experience flavors from their imagination, revolutionizing the culinary and entertainment industries."

Prompt: How to use temp in a prompt

One of the easiest and best ways to do this is using a structured prompt.

▶︎▶︎PROMPT

[goal: Finish this statement.] A dog is to a cat, as a milk bottle is to [Instructions: temperature = .01] (note I used .01 and then 1.0)

Here is a link that shows you the different responses I got. I ran it a few times.

It is interesting to note that with a .01 temperature, the response was looking more at the characteristic of what a cat might be for a dog “chew toy,” “bone”, and yes a “juice box” which is odd but I think it might be looking at how a juice box is crushable.

When I set the temp to 1.0, the responses change to be more about a different type of container and content. Perhaps it was viewing a dog and a cat as similar containers (mammal, animal, four-legged, etc) but with different purposes, instincts, and qualities. The responses were “a carton of almond milk”, “a sports drink bottle”, and “a can of soda.”

To give you a more built-out example, I ran this prompt:

▶︎▶︎PROMPT

[goal: Provide a list of ideas for writing a post about the following - generative ai and lawyers]
temperature = .01]

Here is a link that shows the big difference

News you can Use: You get AI. You get AI.

Lexion + AI

A Wilson Sonsini-backed CLM (contract lifecycle management) provider noted it will leverage its newest investment ($20M Series B) to further enhance its product and AI capabilities.

Aderant + AI

Business management software provider Aderant is rolling out Onyx, an AI-powered outside counsel guideline (OCG) management solution that seeks to set itself apart by automating and validating guideline compliance for law firms.

News you can Lose: 

Prompt engineering has been hyped as the next big (and $$) in job roles. I don’t buy it simply because the tech is moving too fast and baseline prompt engineering will be less needed as the tech and user interface advance.

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