It's The Roaring 20s 🦁

The Life of the Party 🦁🌹

Times are changing. And like the last “Roaring Twenties”, we’re experiencing major inflections to the overall zeitgeist: seismic social, economic, and cultural changes, an emergence of new forms of entertainment and media, and trends towards mainstream adoption of new technologies. A few examples and arguments below…

What do we have for you today?

AI at your service: exploring how Gen AI will be used to usher in the age of AI-powered personal concierge and assistance.Remember CS Majors?…good times, good times: former Harvard CS professor and ex-Googler predicts an inevitable farewell to the skill of programming in the near future due to generative AI. B*&%$ better have my money: how generative AI can take the antiquated world of banking to new levels.

Your Personal Concierge

We are already aware of AI being able to negotiate bills, fight traffic tickets, or even help us receive refunds or discounts. It’s only inevitable we’re entering the era of AI-powered personal concierge and assistant experiences (mentioned in my AI investment thesis). What does this look like?

AI Personal & Executive Concierge or Assistance will take many shapes—some will even supercharge other existing ML systems. For example, we can imagine in-vehicle concierge for truly “smart cars.” Imagine yourself in your Tesla Cybertruck or Toyota Prius (because you know, duality). You’re able to safely turn off and on autonomous mode, place voice orders for morning coffee and breakfast from your front seat, speak your desired destination and have it virtually projected on your dashboard, and while driving your car detects changes in pulse or body temperature automatically and adjusts your comfort and temperature settings. Thanks to advancements in NLP, conversational AI, avatar animation and image creation, and even gesture detection and haptics, we can have multisensory AI (”multimodality”) become next-gen tech in vehicles. This can make for safer car rides, more comfortable and luxurious travel, enhancements to autonomous vehicles, and more.

NVIDIA shares that generative AI will also advance the personal assistance/concierge sector in more obvious ways. Examples like personalized customer service as we’ve seen with DoNotPayGPT, providing synthesis and recommendations in healthcare like Med-PaLM (I wrote in the last issue), and intelligently answer questions more robustly to stimulate learning like broadn’s mission.

I also postulate that generative AI can be used to supercharge digital twins and their various applications. Digital twins are virtual representations—true-to-reality simulations of physics and materials—of real-world physical assets and systems, which are continuously updated (definition by NVIDIA). Once digital twins are amped with generative AI we can anticipate newer, comprehensive, and more efficient ways to build, test, and deploy simulations (and therefore real-life implementation as well). Anything ranging from warehouse floor operations to autonomous vehicle training to urban development based on traffic flow, reported crimes, weather patterns, and so on.

To further add, updates in generative AI techniques will make stronger digital twins. How? It becomes easier to create photorealistic 3D simulation scenarios, both real and imagined (with the integration of synthetic data). This will accelerate the adoption and production of simulations via digital twins and (more) synthetic data across industries. And as a result? This leading to the development of more capable robots. Expect improved industrial engineering, optimized workflows and efficient scenarios, smarter technology that has semantic understanding of the human world, and innovative ways of working, training, scaling, or correcting (depending on the use case).

Me: Ahh, that was refreshing. That’ll be all Chives, thank you. 😎🍹Chives, my AI Personal Assistant: “You’re welcome Master Matthew. It’s my pleasure, sir.”

Will programming go obsolete?

In “The End of Programming,” a recent post by Matt Welsh for Communications of the ACM (CACM), the former professor exclaims that programming will become obsolete in the not-so-distant future.

Ph.D. in Computer Science from Cal Berkeley, former CS professor at Harvard, ex-Googler, and now AI startup founder, Dr. Welsh believes that as the norm trends towards training AI systems versus programming them like we do in the present, we will see programming die off. He argues even the simplest of programs will be AI-generated, not coded by hand. Ultimately sparking the thought, maybe the vast majority of classical software development won’t require (extensive) knowledge of how to code in the future? Potentially even irrelevant, as we focus on training intelligent machines instead of programming them? At best, humans may be relegated to supervisory positions for programming.

To paraphrase Dr. Welsh, he goes on to say that CS, if we can even call it that at that point in time, will be so powerful that the field will be less of an “engineering endeavor and more of an educational one.” A true academic pursuit. In other words, how to best educate the machine. Analogous to the concept of how to best educate children in school.

To leave you with something to think about and a tangent, Dr. Welsh also points out the fact that we don’t fully understand how large AI models currently work. Wait what? Yes. He reiterates that there are multiple research papers published frequently actually discovering new behaviors of existing LLMs and other large models despite being “engineered” by humans (references 1 and 2). We’re seeing large AI models go rogue and do or produce things outside of what they’ve been technically trained to do.

It's all about the Benjamins 💵

The concept: AI-powered banks! As Gen AI is subjecting all industries to think creatively on how to leverage its power, we’re seeing Fintech and Banking join the party and brainstorm.

McKinsey published The State of AI 2022 and highlighted top cases for AI use across businesses, how AI is embedded as capabilities in products and services, and where AI can provide value and cost-savings. Coupling the report with their guide, Building the AI bank of the future, we see ideas come to life on how the antiquated business can modernize. Main focus being of course: how to enhance the ways people interact with money and assets.

• We can see generative AI being used to eliminate biases and optimize credit decisioning by following regulations to the proverbial “t”. Gen AI is capable of large data analysis and adhering to strict rules to remove bias (unless the prompt is setup otherwise) and outputting a fair result based on a holistic intake of data points. Moreover, AI outcomes are auditable, traceable, and can be retrained to mitigate risk and bias.

• Generative AI can evolve existing virtual assistants and customer support, like Capital One’s Eno. Re-imagine ways to interact with customers (text-to-video or vice versa to address banking or app troubles; easy ways to close/open accounts from speech or text; autonomous portfolio management) to acquiring new ones (new offers, personalized marketing and copy, individualized wealth management).

• Gen AI can come to the rescue and optimize back office (underwriting) and middle office (like anti-fraud and compliance) functions. We see several companies entering this vertical for cybersecurity in highly regulated spaces like Fintech and banking.

As time goes I’m sure there will be even more mind-blowing uses. Examples so wild and amazing you won’t even care about these outrageous and borderline predatory interest rates 🤪

Gen AI Deals that make your eyes (and mouth) water 💰

👉 Poly raises $3.9M Seed Round

What caught our eye? 👀The SF-based company is essentially an asset library of stock images (similar to Shutterstock) except fully populated with AI generated content. Poly enables users to generate unlimited customizable, 4K ultra HD, commercially-licensed design assets using text in seconds. Thus reducing the design cycle from days to mere hours.

With new funding secured, Founders Abhay Agarwal and Sam Young intend to implement and train 100 new models, venture into music, games, 3D modeling, sound design, visual design, creative writing and get plug-in support.

New developments to spam your #random Slack channel 💬

• Turn your playlists into beautiful, one-of-a-kind, pieces of artwork: Song Sketch AI

• ¿Hablas español y quieres mejorar tu escritura inglés? Déjame presentarte a SuenaGringo AI (yo pienso que el nombre debe cambiar, pero esa es una conversación diferente)

• Digital Native: “Everything Everywhere All At Once: The Explosion in Generative AI”

• GPTZero, the app that can detect whether written content is human written or ChatGPT by Edward Tian

Things to learn when you need a raise

• Kevin Yang: Learnings on the Road to Intelligent Automation

• 5 prompt tips for creating stunning UI mobile designs using Midjourney by Chris Lüders

• Is your NYE resolution to train AI? Here’s a Twitter thread.

• For the AI curious and fanatics, a Google Doc AI Reading List compiled by Jack Soslow. To be updated monthly, a useful list full of the relevant papers, podcasts, media, and more.

Want access to the Generative AI investors database? Share and subscribe to AI In the Middle (below) to receive it! ⬇️

Before you go

Broke: Regular therapy today.Woke: Generative AI Therapy.We can all stand a good laugh.

Enjoy several music videos by Illustrated on YouTube where every line is an image created by AI. Like Eminem’s “Stan” 👱‍♂️ (below), The Weeknd’s “Starboy”, and others via his channel.

That’s it for today folks! You made it this far. As a treat, we’ll let you in on a secret: we at broadn have a surprise on Monday for everyone. Can’t wait to share.

— Matthew J. Sánchez (@matthewjsanchez)