GenAI Music Creators Udio and Suno Go Viral, But Instigate Copyright Concerns (Ex-Music Industry Alum Thoughts)

The Week's AI Advancements: Udio and Suno AI Music Generators, Humane AI Pin, Spotify AI Playlists, HeyGen, Tesla's Robotaxi, Google CloudNext Conference, Axion, and a review of Motion AI.

Watch the video de-brief if you don’t have time to read.

Advancement #1: Udio and Suno GenAI Music Generators Crank it Up

This week has been a big week for GenAI music generators. First, Suno AI started to pick up significant traction, creating 2 minutes songs (both melody and lyrics) from prompts.

Then Udio, developed by ex-Google DeepMind researchers was launched on Wednesday (April 10), elevating functionality by adding a remix option that lets users edit their tracks using text descriptors, effectively transforming everyday music fans into skilled producers (but I dealt with crazy website issues… like insane bugs. Who did the QA testing?).

Leading the investment in Udio are Andreessen Horowitz (a16z), a venture capital firm based in Menlo Park, California with $35 billion in assets under management. Angel investors include Mike Krieger (co-founder and CTO of Instagram), Oriol Vinyals (head of Gemini at Google), and musician Will.i.am.

Both apps are designed to simplify music creation, allowing users to generate a mastered track in under a minute, specify genre, add specific subjects or personalized lyrics, and naming inspirational artists.  

My Initial Thoughts: Hey, come on. You know I’m going to take this time to talk about music copyright. After all, it was my life for a good period of time.

I have talked about this debate every opportunity I have, as this is an ongoing question that has not been answered by any nation in the world yet (I talk about the copyright question and my thoughts very heavily in my uDemy course, if you want to expand on it).

My opinion is this, and I believe it’s also very well informed (because I’ve studied music copyright law for many years). I lean towards believing AI music generators in themselves are not copyright infringement. Because GenAI models create entirely new content, it is statistically impossible to be close to the original.

However, what I strongly believe will be stipulated as copyright infringement in future laws is using an artist’s likeness or brand to create music in their style. Similarly, using an actor’s face to recreate movies in their likeness, for example: should not be allowed.

Less clear: using copyrighted material to train these models. I do see the argument for Fair Use exemptions (since it’s nothing like the original), but also think the laws are outdated and future iterations should require licensing permission. Sam Altman disagrees though, and thinks the laws should consider whether or not that copyrighted content is publicly used or distributed versus privately used (which is a distinction that has never been considered before in copyright). Andreessen Horowitz also is a strong voice saying it’s Fair Use.

What this all means? There’s a legal loophole right now, but future laws will adjust.

Advancement #2: Humane AI Pin Becomes Publicly Available This Week

The Humane AI Pin launched this week, a new device hoping to instigate a post-smartphone future. It’s a screenless phone worn on the chest designed to combat phone addiction and usher in a new era of personal devices. Described by the company as your "second brain," this device uses voice commands, tapping gestures, and body projections—typically on the hand—to operate. Along with its AI OS, Cosmos, the AI Pin represents the dawn of ambient computing.

The AI Pin (demo video here) has gotten overwhelmingly negative reviews that got almost nothing right in terms of solving functionality for cell phone users. Voice commands would often fail to execute, the cellular connection is only through t-mobile with no connection to your existing number, many felt the projector was invisible in some light and the Pin even refused to translate things properly, which was a major selling point used in the marketing campaigns. Users reported that rather than translated things, it would just repeat it back in a horribly and occasionally offensive accent.

Starting at $699 with a required subscription of $24 per month, this is clearly a first-generation prototype not worth the time or money.

My Initial Thoughts: I saw the Humane Pin last week in action (before launch) at a coffee shop in SF by one of the developers. Everyone chattered as we saw the green projection and he closed and opened his palm and tapped his chest. It was definitely a cool device to see, but only because I’ve never seen anything like it before, not because it was actually useful. As is, its a combination of a wearble Siri button that doesn’t actually listen to you, with a camera and a built-in projector that only shows basic its it’s basically a combination of a wearable Siri button with a camera and built-in projector.

Sounds… exciting?

Advancement #3: Speed Round

Dozens and dozens of mini updates these week. Not worth expanding on, but worth you knowing about them.

SPOTIFY AI PLAYLISTS

Spotify introduced into beta Monday AI Playlists, which allows users to generate a playlist based on written prompts. The feature will initially become available on U.K. and Australia.

HEYGEN RAISING ROUND VALUES COMPANY AT $500 MILLION

HeyGen, an AI video generation startup is closing a $60 million funding round led by Benchmark, valuing the company at $500 million. HeyGen has experienced a 20x increase in revenue from the previous year, but still remains second in the market being Synthesia evaluated at 1 billion.

GOOGLE CLOUD NEXT CONFERENCE RELEASES

At Google’s Cloud Next Conference this week, they launched Gemini Code Assist, which is similar to Github Copilot, came out with an enhanced image-generating tool called Imagen 2 that can be accessed through Vertex AI for developers that can do text-to-live-images, as well as announced their first custom-built Arm processor called Axion, allegedly 30% better in performance than other Arm-based instances from competitors like AWS and Microsoft.

TESLA’S ROBOTAXI

Elon Musk tweeted that Tesla will be announcing its Robotaxi on August 8 with an image teasing how it will look. However, regulators in many states (including California) said Tesla hasn’t applied for permits it would need.

Review: Tried out Motion AI

This is not something I typically do, but since I’m trying out AI tools alot, I figured if there was a review that would be helpful I should share it. I have seen advertisements for this amazing tool for months called Motion AI and it’s gone viral. It’s supposed to help you organize the long list of tasks and optimize your schedule, leveraging AI to proactively schedule your tasks per day and save you 30% time.

I’m slow to sign up for stuff if I ever have to enter a credit card, but since there were 3 nights this week that I got less than 2 hours of sleep, I figured maybe it could help me. I have to admit I still primarily work off of Apple Notes or Google Keep the majority of time. XD

But… unfortunately, I hate it for the following reasons:

1. Too many fields to create tasks. I don’t want to add a title, start date, deadline date, whether it’s a hard deadline, description, etc. I need to type it quickly, say how long it’s going to take, and move on. Spending a minute to enter a single task is insanely annoying for someone moving 1,000 miles per second.

2. Many projects/tasks require a sequential order, and Motion AI just cannot handle it. It proactively schedules things out of order with “AI”, but when you try to adjust based off the calendar and the execution order that you are aware of offline, the system takes it as a manual override and “locks” that task. IE, it is now excluded from the AI scheduling as you continue to move things around.

3. A lot of bugs, honestly, with basic field inputs and even the calendar.

4. You overall cannot move anything around and have things shift around it like the demo shows. Any intervention by you as the user disables the AI for that task. That basically renders 80% of my tasks useless.

I tried to use it for a solid 5 hours and honestly, I found myself back in Notes. So long story short… not there yet. Fantastic idea, and I can’t wait for it to work, but don’t get lured into the hype.