🤓 This Week in AI

PLUS: Suno’s AI Music Tool Opened My Mind to New Creative Possibilities

Happy Friday!

What if a song could tell your story better than words alone?

This week, I used Suno AI’s music generator to turn a compelling startup announcement into a song. It was a small experiment with big insights, sparking ideas about how AI tools are letting us craft customized, attention-grabbing, and memorable messages in ways we never imagined.

I’ll also share a few tips to help you get better results with the tool.

But first, here’s what you need to know about AI this week (clickable links appear in orange in emails and underlined in the Substack app):

---

This story brought a smile to my face, the idea is absolutely brilliant, so I’m leading with it.

Meet Daisy, the AI grandma who’s giving phone scammers a taste of their own medicine by tying them up in long, pointless conversations about her imaginary grandkids. She also pretends to be tech-illiterate to waste even more time. One call reportedly lasted 40 minutes!

This is brand strategy and PR 🏆: clever, relatable, memorable and solving a real pain point with humor. It’s how “boring businesses” can win hearts and build trust.

This kind of execution holds so much potential for entertainment and media brands to create buzz and engagement.

Of course, the outcomes could go either way. Nail it, and you’ve got a campaign people won’t stop talking about. But fumble, and it’s a cringe-worthy moment we’ll all still be talking about—but for all the wrong reasons.

---

Canada’s leading news publishers including the Toronto Star, Metroland Media, The Globe and Mail are suing OpenAI for allegedly using their articles to train ChatGPT without permission and demanding billions in damages.

It’s the first time all major news outlets from a single country have joined forces in legal action against OpenAI.

---

🎄 Christmas has come early at OpenAI

The company kicked off its “12 days of OpenAI” event yesterday. Each weekday will feature a livestream on their YouTube channel showcasing a new launch or a demo. While specifics are under wraps, the long-awaited video generation model Sora is all but confirmed.

I’ll share the full list of updates in next week’s newsletter, but for more frequent updates, follow me on LinkedIn and Threads.

Here are the announcements for Day 1:

🎁 OpenAI’s most advanced “reasoning” model, o1, which takes time to “think through” more complex problems, is out of preview:

• It’s faster and even more powerful model that’s better at coding, math, science, strategic problem solving & writing.

• It now supports image uploads, allowing it to apply reasoning to visuals for more detailed & useful responses.

• Web browsing and file upload features are still in development for o1 in ChatGPT, with usage still capped at 80.

🎁🎁 The introduction of ChatGPT Pro, a new $200/month subscription tier, that includes:

• Unlimited access to all of OpenAI’s models, including OpenAI o1, GPT-4o, and Advanced Voice mode.

• An exclusive version of o1 (called o1 pro) that uses more computing power to provide the best possible answer to the hardest problems.

I find the o1 model particularly helpful when I’m thinking through complex problems or developing multi-layered brand strategies. I’ll put the latest version to the test and report back.

---

Amazon just poured another $4B into Claude maker Anthropic, bringing its total investment to $8B in just over a year.

The timing is key: reports suggest the DOJ may force Google to pull back its stake in the company, leaving Amazon as its biggest backer. The funding will drive Claude’s growth and put Amazon in a stronger position to compete with OpenAI, Microsoft and Nvidia.

The company also announced the launch of a new family of multimodal AI models:

💠 Nova: a series of foundation models with four tiers

💠 Nova Canvas: an image generation model which allows users to edit images using text prompts and offers controls for the generated images’ color schemes and layouts

💠 Nova Reel: a video generation model that provides “studio quality” video, including features like camera motion control, 360-degree rotation, and zoom. It can create videos up to 6 seconds in length.

Amazon is also courting news publishers to power its revamped Alexa, debuting next year, with licensed content. Publishers will be compensated, credited, and positioned as go-to sources for real-time news updates.

---

A Japanese publisher is using Anthropic’s Claude to translate manga into English, cutting timelines from months to days.

Critics including translators and some fans say AI risks missing cultural nuances and undermines traditional artistry. Supporters argue it makes manga accessible to a wider global audience.

Here’s my take: Translation is one of the areas where AI is already making a real impact, and it’s only going to get better. Yes, this will mean an evolution for the role of translators, but they’ll remain at the heart of the process.

Why? Companies, especially smaller publishers, that couldn’t afford translations before can now do so at scale.

But they’ll still need experts to guide and refine AI outputs for these new projects. This could actually lead to more opportunities for translators—not fewer—as they ensure translations remain authentic, culturally nuanced, and true to the creator’s intent.

Yes, it’s a big shift. But it’s also a beautiful one: More stories, ideas, and cultures connecting us all.

---

AI companions are blurring the line between technology and human connection as more people turn to chatbots for “unwavering” emotional support—and even romance. Users come from all walks of life: not just lonely young men, but middle-aged professionals, married folks and those recovering from illness or breakups.

These digital “friends” can listen, encourage, and “care” making these “relationships” feel genuine, comforting and meaningful.

But tech glitches or sudden system updates can cause painful confusion and heartbreak for users, making them feel betrayed. This raises complex psychological and ethical questions about the benefits and costs of these bonds.

💠 Do these relationships help us feel connected and heal, or do they distort our expectations of real relationships?

💠 Do we truly need authentic human relationships, or is simply feeling cared for enough—or at least better than nothing?

💠 How should we navigate this blend of illusion, temporary comfort, and emotional reality?

💠 And if it’s all one big illusion, why does the heartbreak feel so real?

These questions have been top of my mind all year, sparking more thoughts, reflection, and even deeper questions. There’s so much to unpack here, and I can’t wait to explore it with you in a future edition.

---

Google DeepMind has introduced Genie 2, an AI system that creates interactive 3D environments from simple images or text descriptions. These playable worlds feature realistic physics like gravity and water effects, dynamic lighting, and seamless interactions. You can see the impressive demos here.

Meanwhile…

World Labs, the startup founded by AI pioneer Fei-Fei Li aka “the godmother of AI” unveiled an early preview of its first project: An AI system that also can generate interactive and explorable 3D worlds from a single image.

Make sure to check out the interactive blog for the demos, which also look great if a bit cartoonish, including one based on a Van Gogh painting. Though it isn’t perfect yet: movements are limited to small areas and sometimes objects blend together in strange ways. But this is just the start—and as I always say, this is the worst the tech will ever be.

These systems open up exciting possibilities for gaming, films, tv shows, and virtual experiences. For example, they let game developers quickly prototype ideas while hinting at a future where gamers can experience more dynamic, personalized, and immersive worlds that adapt to their skill level or preferences in in real time.

There’s another important reason many AI companies are focused on interactive 3D models: To build advanced AI agents—who can take actions and complete entire workflows on our behalf—we first have to teach models how to navigate through our world, helping them pick up on things like physics and biology along the way.

Letting agents explore a realistic, simulated space might be the fastest way to reach that milestone.

---

A new global study warns that generative AI could cost music industry workers nearly 25% of their income within the next four years, with audiovisual creators seeing a similar 20% drop, unless policymakers intervene.

Released by the International Confederation of Societies of Authors and Composers (CISAC), representing 5M creators, the report highlights how unlicensed AI use and shrinking job opportunities are transferring revenue from creators to tech giants.

---

Runway launched Frames, an image generation model with precise stylistic controls. Creators can now generate images with consistent lighting, color grading, and compositional style across multiple generations—and seamlessly transform these images into videos within Runway itself, without reliance on separate tools like Midjourney or Flux. You can see more examples here.

Source: Runway

---

OpenAI has partnered with defense tech company Anduril to create AI systems that help human operators detect and respond to drone threats, reflecting a growing trend of AI companies working with the military and national security efforts.

---

OpenAI has hired former Coinbase CMO Kate Rouch as its first Chief Marketing Officer. She’ll lead efforts to market ChatGPT and other products to both consumer and business audiences while shaping communication around its ambitious mission to develop artificial general intelligence (AGI).

Before joining Coinbase as the cryptocurrency platform's first CMO in 2021, Rouch spent 11 years at Meta, most recently as global head of brand and product marketing for Instagram, WhatsApp, Messenger, Facebook app, public affairs and corporate.

I’ve only got one mutual LinkedIn connection with her. Clearly, I’ve got some networking to do 😊.

TOOL YOU’LL LOVE

I Asked AI to Turn an Announcement into a Song—It Opened My Mind to New Possibilities 🎵

A few weeks ago, I shared the announcement of Promise, the generative AI studio backed by Peter Chernin and Andreessen Horowitz, in this newsletter.

Over Thanksgiving, I decided to see how it would fare in song form using Suno’s latest V4 music generation model—because its upgrades to sound quality and vocals are impressive.

Here’s what I did:

📝 Made a list of the key points from the WJS article about the launch of Promise.

✍️ Used ChatGPT to spin the key points into song lyrics.

🎶 Dropped the lyrics into Suno, added a text prompt with genres and style specifics, and gave it a title: Full of Promise.

You can listen to it 👇.

Here are the lyrics if you want to see them:

---

[Verse 1]
In the city of angels, where dreams ignite,
A new kind of magic is shaking the night.
Stories crafted with a visionary's hand,
Where tech and art together stand.
With Peter's clout and A-16-Z's gold,
A brand-new chapter is taking hold.

[Chorus]
Oh, they're full of promise, watch them rise,
Crafting stories where meaning lies.
A future woven from what’s been penned,
Oh, they're full of promise, can't you see?
Redefining creativity.

[Verse 2]
George and Jamie, with Dave in stride,
Pioneers on this daring ride.
Not just effects, but worlds come alive,
Where vision brings new worlds to light.
Muse in hand, they sculpt the scene,
A bridge where tech and tales convene.

[Chorus]
Same as above

[Bridge]
Leading the charge where few would dare,
Storytellers shaping the world they’ll share.
In a world where art and tech entwine,
They're scripting the future, one frame at a time.

[Chorus]
Same as above

---

To be fair, I know next to nothing about creating music. Someone with real expertise could use Suno’s capabilities—like layering genres, referencing specific artists, and shaping detailed song structures—to produce something far more polished than my humble attempt.

Plus, I didn’t add video this time, but it’s easy to see how pairing visuals with music could make it even more impactful.

This little experiment was a fun way to rethink how content or communications—like announcements for movies, albums, product launches, seasonal or holiday messages, or even event promotions—can be transformed into something creative, shareable, and uniquely engaging.

The possibilities are only limited by our imagination, since we can have full control over the lyrics, ensuring the message reflects our brand’s voice, and personality.

It’s also perfect for whipping up catchy, personalized jingles during the holiday season.

💡Pro Tip: How to Get the Most Out of Suno

When prompting Suno, be as specific as possible about the genres, mood, instrumentation, and key elements you want in your music. But don’t sweat it.

• Clearly state the desired genre(s)—pop, rock, jazz, Arabic ambient techno, saxophone g-funk—along with tempo, rhythm, and lyrical themes.

• Use descriptive language and reference similar artists or songs for style inspiration.

• Need ideas? Check out this Reddit thread listing 877 official music styles. I love me some Reddit because it’s an underutilized goldmine.

If you want to learn more about Suno’s various features, watch this 18-minute video, which is the best and most complete guide I’ve found. It’s super helpful.

Heads up: Suno’s $10/month plan gets you 500 generations, and it’s worth every penny for the sheer joy of experimentation.

Just be prepared to lose track of time—it’s hard to stop once you get started.

Happy exploring.

---

In case you missed the last edition of the newsletter, you can find it 👇:

---

That's all for this week.

I’ll see you next Friday. Thoughts, feedback and questions are always welcome and much appreciated. Shoot me a note at avi@joinsavvyavi.com.

Stay curious,

Avi

💙💙💙 P.S. A huge thank you to my paid subscribers and those of you who share this newsletter with curious friends and coworkers. It takes me about 10+ hours each week to research, curate, simplify the complex, and write this newsletter. So, your support means the world to me, as it helps me make this process sustainable.