AI with ALLIE

The professional’s guide to quick AI bites for your personal life, work life, and beyond.

October 31, 2023

Riding the waves of my newsletter journey, edition two is here.

This time, we’re chatting about Spotify's Daylist, what I think it says about the consumer’s role in AI personalization, my prediction for the future, and the pace of change in AI. And, of course, I won't let you leave without sharing a curated list of AI tools, courses, and intriguing reads that have recently piqued my interest.

I'm also thrilled to share the resounding success of my first course, offering a glimpse into what went behind the scenes and what we're thinking for the future (of course, you will all be the first to know).

Let's dive in.

Peeling Back the Algorithm

I remember a day when Netflix was broken.

It was years ago.

Netflix had accidentally exposed its “secret” movie and TV categories. I’m not talking about the 100ish content codes that are plastered all over the internet, this was a whole other level. Instead of a standard category name like “rom-coms”, Netflix’s homepage was unknowingly displaying crazy specific headers like “1980s ballad-filled powerful woman romantic comedies”. The Atlantic reported in 2014 that Netflix had almost 77,000 categorized subgenres, and I suspect those were the headers we were seeing. It was…in a word…awesome. It gave users a glimpse into the hyper-personalized world that AI sees behind the scenes.

Then Netflix was fixed, and it felt like my perfect little ice cream sundae was melting. I had seen the underbelly of the Netflix algorithm, I had gotten an inside look into my own heart, and then I was whiplashed back into the all-too-familiar and all-too-boring territory of normal bland headers like 'rom-com' and 'sci-fi thrillers'.

The new “Spotify Daylist” feature feels oddly reminiscent of the bug that everyone wanted to keep from Netflix.

And I think it’s actually an omen.

What is it: 

Spotify Daylist offers us real, personalized insights into the intricacies of our musical tastes and patterns. It’s a journal-like playlist that’s tailored to every twist and turn of our day, powered by AI. Think of it as the sophisticated older sibling of the random Netflix categories we once saw.

It evolves with your mood, day's tempo, and even climatic shifts. It breaks down the ambiguity of broad genres and gets into the weeds. So, instead of just a ‘rock’ tag, you might find get served up a ‘sunset grunge-rock on a rainy day’ playlist (or, in my case, ‘northern soul disco fever night’ and ‘dirty pop throwback Monday morning’).

It changes multiple times a day. If you just open the Spotify app, type in “daylist”, and hit ‘enter’ — you’ll see what I mean.

Some other examples from my friends playlists:

• golden age film score monday evening

• energy viral monday afternoon

• original broadway cast thespian monday morning

• mountain music stomp and holler monday morning

• sad emotional monday morning

• city pop fusion monday evening

It’s literally an AI mood ring, but for music.

Why this matters: 

Algorithms try to uncover (or, more accurately, exploit) what we are interested in, shape our preferences, and guide us into categories. They do it to sell ads (like Twitter), they do it to sell products (like Amazon), and they do it to keep your attention (like TikTok).

Now, we can say that this is a value-add for the end customer. Perhaps you are happier when you see ads meant for you, shirts you might like, and content that makes you laugh. But it’s abstracted. You’re not told why that ad is there, or why that shirt is there, or why that video is there. You’re just shown the thing, and you hopefully like the thing.

And these companies do it without sharing that transparency, that value, that insight back with us. There are larger legislation conversations about AI explainability, but I think what the vast majority of people don’t realize is that explainability is not only valuable for risk mitigation, it’s valuable for self-reflection.

It’s not just a defense move that reduces harm, it can be an offense move that grows revenue. Explainability can switch the “give and take” model more in the user’s favor.

You see, the beauty of Spotify Daylist isn’t just the hyper-personalized recommendations. I could do that with a burned CD in the 90s. It’s taking personalization one step further with a transparent, share-worthy approach. They literally tell you the crazy specific Netflix-like header, plus the full explanation. Just look at this example below.

It’s educating you on your preferences and making it fun. You’re getting a front-row seat to your own evolving tastes and inclinations. It’s not just a playlist; it’s a lesson in self-awareness.

Despite this being a media example, this doesn’t stop at telling me to listen to Paint it Black instead of Rolling in the Deep (and honestly, they’re both great, who am I kidding).

AI has the opportunity to not only deliver a hyper-personalized, goal-oriented, and value-oriented experience, but it also has the opportunity to be transparent in that experience, and to give that valuable insight back to the customer and actually help them learn about themselves.

The genius here is that every technology company mines a lot of data, but Spotify actually gave it back to the user. And I think we’ll see more examples of this in the coming year.

So what data are you holding on to? And what parts can meaningfully shape your customers’ experiences for the better?

Let’s talk about it.

What your business should do:

1. Embrace Transparency: As business leaders, integrating transparency in AI applications can heighten user engagement and trust. At a basic level, for example, think about platforms that explain why certain ads are shown or why specific content is recommended — this doesn't just satisfy user curiosity but also creates a feeling of control and mutual respect. And at an advanced level, think about showing credit card options to customers and explaining why these two should be their top picks — it also makes their decision faster. By transforming data into a two-way conversation, businesses can create a bond of trust, increasing long-term user loyalty and customer lifetime value (CLV).

2. Innovate with Personalization: The success of personalized recommendation engines, which curate items tailored to individual tastes, is no coincidence. What if, beyond just delivering tailored shopping suggestions, you were given insights on why each item was chosen, based your past preferences? For example, you’re shopping for a home, and the empty rooms are designed by generative AI to mimic your wants and style and writes up a story about how you would enjoy living there and why. By showcasing users' journeys and habits back to them in a tangible, fun manner, companies can provide a mirror for self-discovery, enhancing the value proposition of their products.

3. Think Meme-ability: In the digital world, going viral is gold. Features that entertain and are created with social sharing in mind can catapult a brand or feature into internet fame overnight. Think about Buzzfeed quizzes that we’re all guilty of clicking on, or zodiac forecasts that tell us our future. Create features that not only serve but entertain and engage in a socially shareable way. Memes are the language of the digital age and integrating meme-ability or emotion-evoking shareables can exponentially increase your feature's reach and resonance, even if it’s just a quirky playlist title.

In a world where we're all hunting for personalization—from our news to our coffee orders—it's fascinating to see technology not only meeting but exceeding our desires and helping us in the process. Whether it’s a “caffeinated morning pop” or “throwback guilty pleasure Saturday night,” may your Daylist always resonate with your heart’s current tune. (And if not, just listen to more of what you do want and let the product catch up to you.)

The Speed of AI and Pace of Change

Normally, I would only have one “big” story in these newsletters, but I really want to talk about one other topic. And that is change.

Change is not only constant—change itself is changing. The pace of change is ever increasing, and all businesses, organizations, people, teams, countries, should be prepping for one thing: more change, and with it, even more “VUCA” (volatility, uncertainty, complexity, ambiguity).

A sobering thought on the speed of AI: something that took 50 engineers, 4 product managers, and 1.5 designers 6-12 months to build in 2018/2019 can now be (mostly) done by two people in a few days.

And I know this because I led the team in 2018. And a 2-person team pitched me on the same idea a few weeks ago.

Gulp.

You might scoff and think we had a “bloated” team, but you’d be (mostly) wrong. A lot of folks, especially those new to AI, may not appreciate how far we’ve come since 2018 when it comes to natural language processing (NLP) and document parsing (the ability to take a scanned PDF and turn it into something computer-readable). We had dozens of engineers solving a document parsing problem that is now 95% solved, extremely cheap, and accessible via API.

Let me jog your memory with some AI releases since that project:

• spaCy got non-English languages late 2017

• BERT wasn’t published until October 2018

• Textract (Amazon’s document parsing product) didn’t drop until November 2018 and didn’t GA until May 2019

• And GCP’s Textract competitor, DocAI, didn’t come out until the following year in November 2020

That’s right: Google’s PDF reader didn’t exist until 3 years ago. And now, only 3 years later, you can have a cute little “chat” with a PDF using natural language in ChatGPT (released Oct 30, 2023). You can ask it questions, get a summary of it, ask it to write a follow-up, and more. In less than the time it takes to get a degree, the entire landscape of data access has changed. What the heck.

Whatever 3-year plan you have for your business is sure to be shaken up.

So don’t just expect change, build a business that relies on it.

Missed my AI for Business Leaders course?

Our inaugural cohort had hundreds of dedicated students and an astounding average rating of 9.5 out of 10 stars. And little known fun fact: I wrote the entire course outline with my thumbs on my iPhone while on safari with no internet.

Over the span of two months, my team and I created, promoted, launched, and ran the entire 2-week class. The cohort included top executives and business leaders from CVS, Mastercard, Amazon, Deloitte, General Motors, Mattel, Doordash, Cisco, Autodesk, KPMG, Twilio, Nokia, Heineken, and more. And it was a blast. My inbox is filled with success stories, and I couldn’t be happier.

While my team and I are strategizing the best way to scale the course, I'd love for you to stay in the loop. So if you’re interested in joining and getting your AI groove on, secure your spot in the potential next cohort by signing up for the waitlist here.

Tools, courses, and blogs that caught my eye:

Over the last few weeks, I’ve pulled together some of the top releases, and my take on each one. Check it out.

1. Spotify AI Voice Translation - Breaking down linguistic barriers, one podcast at a time. Spotify's new AI-driven voice translation lets listeners enjoy podcasts in multiple languages, all while preserving the podcaster's unique voice and style. Another leap for global connectivity and making content universally relatable. I guess I should do more podcasts next year? (read it) (my thoughts)

2. ChatGPT's Sensory Upgrade - ChatGPT now integrates voice and image recognition, enabling dynamic conversations and image-driven interactions. Users can opt for voice chats, use photos for context, and select from five distinct voice options. I can’t work out of a coffee shop because I am non-stop voice dictating to ChatGPT and other AI systems (read it) (my thoughts)

3. Amazon's $4B Investment in Anthropic - Amazon plans to invest “up to $4 billion” in Anthropic, aiming to synergize Anthropic's AI safety research with Amazon Web Services (AWS) infrastructure. As part of the deal, Anthropic's advanced AI, Claude, will see expanded support on Amazon Bedrock, while AWS becomes Anthropic's primary cloud provider. I see it as a hardware play. Google also invested $2B into Anthropic just this week, but because it was 2 billy less than Amazon, no one is talking about it (read it) (my thoughts)

4. DALL·E 3 - OpenAI just released their third generation for DALL·E with improved image generation and is currently available to a limited group, with wider availability for ChatGPT Plus and Enterprise users scheduled for October, expect even more multi-modality (read it) (my thoughts)

5. Full Stack LLM Bootcamp - Hey devs! A team of UC Berkeley PhD Alumni with years of industry experience is offering an LLM-specific training program for those with experience in Python. It’s 8 lectures, 100% free, and covers the full stack from prompt engineering to user-centered design. I love the team behind this one (sign up)

6. Adobe Project Primrose - Adobe MAX 2023 had tons of incredible announcements, but what wasn’t on my 2023 Bingo card was Adobe getting into fashion, or anything tactile really. Project Primrose is a smart fabric that allows for reflective light-diffusing modules that are low power, modular, interactive, and flexible. Designers can use these modules on furniture, handbags, billboards, and apparel. Adobe is really rising to the top of these AI players (see it) (my thoughts)

7. Google Search Generation Experience - I’m sorry but I have to say it: what an awful name. That said, Google is expanding to text-to-image generation for those opted-in to Google’s SGE via Search Labs. Similar to Midjourney or DALL-E, you can type into Google search bar “draw an image of…” and generate a few images based on your prompt (sign up) (my thoughts)

8. Meta AI Avatars - Across WhatsApp, Messenger, and Instagram, Meta is releasing AI characters for everyone to chat with. These avatars are collaborating with big name influencers to make the interaction more enticing, including Billie (aka Kendall Jenner), Victor (aka Dwyane Wade), Tamika (aka Naomi Osaka), and Dungeon Master (aka Snoop Dogg), but honestly, their approach is a bit odd (read it) (my thoughts)

Feedback is a Gift

I would love to know what you thought of this newsletter and any feedback you have for me. Did you have a favorite part? Wish I would change something? Felt confused? Please reply and share your thoughts or just take the poll below so I can continue to improve and deliver value for you all.