50 AI Things
Welcome to People vs Algorithms #50.
I look for patterns in media, business and culture. My POV is informed by 30 years of leadership in media and advertising businesses, most recently as global President of Hearst Magazines, one of the largest publishers in the world.
A bad week for billionaires. Listen to the new podcast here.
Brian Morrissey and I break down why very smart and successful people like Sam Bankman-Fried and Elon Musk are stumbling so badly.
Unintended consequences
This is the thing about technology. While solving one problem, it creates new ones. The faster and more profound the technological change, the more difficult it is to process its second order consequences. Social media has proven to be that kind of change. Crypto too. AI is coming next.
AI’s impact will be even more profound and its unintended consequences vast, simply because it will force change at the atomic unit of human labor, across every industry. The world is about to be rationalized around a technology that forces the price of things many people do to approach zero.
I started this note intending to catalog the avalanche of new AI technologies in order to put some shape to what’s ahead. I will get to that in a sec.
The world kept pulling me back to the inflection point that is this moment, and how lucky I feel to witness both sides of mountains of change, the run up of IP based technology and its aftermath, the evolution of basic digital publishing to forums to social networking then social media, change so profound it’s swallowed the old, from publishing to TV, advertising to commerce, step by step by step.
This week showed more unintended consequences of crypto change. The FTX meltdown is no doubt the result of a regulatory system incapable of keeping up with the torrid pace of innovation, and the freewheeling antics of crypto's genius protagonists like Sam Bankman-Fried (SBF), egged on by a culture that worships capitalist cowboys. This is the soul of America’s dominant innovation engine, but it’s prone to messy corrections. The unfolding saga will cast a dark shadow on crypto, defi and blockchain—another set back on its long arc of adoption. I suspect it will delay but will not suffocate the disruptive promise of Bitcoin as decentralized currency, of Ethereum as a mechanism to efficiently manage complex interactions between unrelated parties or as the blockchain broadly as a borderless global system of record for everything including what comes next in social networking. Technology is a continual ebb and flow. It breaks shit. Sadly, people get burned. We lurch forward. This one will take longer than some hoped.
Twitter’s become a dumpster fire fueled by unintended, if predictable consequences of something else. Hubris. This genius let the dopamine get the best of him and made a very, very expensive impulse purchase. Admittedly, the timing was bad. That’s forgivable. But, anybody who has ever run a media company, especially an ad supported one, could have provided infinitely better counsel on how to proceed after taking over. Dumb, arrogant mistakes. Recruit better advisors. Plan the transition with more sensitivity. Never forget lessons from high school—no good party ever went down without the libs. Even the jocks aren’t going to the Maga mixer. Elon… you need these people. This is media, not a rocket company. Avoidable, unintended consequences.
You are not your audience
Today we sit in between the old social media and the next. Ian Bogost over at the Atlantic provocatively suggests social media's time is up; its foundational sickness is idea that everybody needs an audience, all the time. I kinda think he's right. From "The Age of Social Media Is Ending":
That was a terrible idea. As I’ve written before on this subject, people just aren’t meant to talk to one another this much. They shouldn’t have that much to say, they shouldn’t expect to receive such a large audience for that expression, and they shouldn’t suppose a right to comment or rejoinder for every thought or notion either. From being asked to review every product you buy to believing that every tweet or Instagram image warrants likes or comments or follows, social media produced a positively unhinged, sociopathic rendition of human sociality. That’s no surprise, I guess, given that the model was forged in the fires of Big Tech companies such as Facebook, where sociopathy is a design philosophy.
Someone asked me at a party last night what comes next. Tough question. I suspect the next wave will be found in smaller, more intimate social groups, built around the trust and familiarity of friends and family, more purposeful communities assembled around passion points and professional interests. Social networks are fragmenting. Broad horizontal networks like FB, Insta and Twitter will persist but without their hegemony. Substack is an interesting place to look. Watch how new engagement tools take shape around content communities and author creators. The next media manifestation will emerge around the magnetic power of people brands, not institutions.
A fortunate unintended consequence… our current travails will provide fodder for streaming storytellers for years to come. I can’t wait to watch the Musk / Twitter story or the SBF biopic. Macaulay Culkin will bulk up and play Elon. Gaten Matarazzo could play Jason Calacanis but he’s too young. Instead, AI will stand in, trained on Jason’s endless library of podcasts and YouTube videos, indistinguishable and perhaps slightly more likeable than the real thing.
The next will be more unpredictable
Meanwhile, we have found a new frontier to project our aspirations and capital upon, just in time for another crypto setback. It is clearly AI. This wave’s unintended consequences will be even harder to decipher.
By now most of us have read the AI generated story masquerading as a real human-made story, ending with the author revealing how the AI is almost as good as the real thing but without the intentionality or nuance or soul. It's a journalistic Turing Test and what it reveals is that large language models and diffusion generators are remarkably good at turning data to patterns to things that mimic human intelligence. Almost. For now we take comfort in the realization that these are tools not replacements.
Some manifestations are more threatening than others. Artists are obviously not happy about the unintended consequence of AI making them unpaid raw materials for diffusion robots (a good article here about how illustrator Hollie Mergert felt about a model trained on her corpus of work).
A recent experiment in the New York Times found NYT Food reporter Priya Krishna preparing an entire Thanksgiving menu from AI. With some massaging, GPT-3 spit out an admirable Thanksgiving ensemble.
Minutes later, I had a complete menu that seemed both plausible and intriguing: pumpkin spice chaat, green beans with miso and sesame seeds, naan stuffing, roasted turkey with a soy-ginger glaze, cranberry sauce that’s not too sweet and a little spiced (yes, that’s the full recipe name) and pumpkin spice cake with orange cream cheese frosting.
Unsurprisingly, the result was pretty much a flop.
The cake was dense and more savory than sweet. The naan stuffing tasted like a chana masala and a fruitcake that had gotten into a bar fight. The roast turkey recipe called for a single garlic clove to season a 12-pound bird, and no butter or oil; the result was dry and flavorless.
Humans are pretty tough to compete with it turns out. It's hard to match raw processing power and pattern recognition with an ability to smell, to taste, to feel, to remember. "The individual behind the dish is just as important as the recipe itself," one of the researchers notes. Ambitious recipe generation is perhaps the ultimate Turing challenge. No, AI is not going to take the job of recipe developers any time soon. Nor will it replace a writer with something to say. Intentionality is important.
Certainly as a tool for inspiration, a friendly assistant to retrieve or sort information, deliver perfunctory instructions on say, how to boil the perfect hard boiled egg, AI has much to offer today. But It was hard to predict how a Usenet forum would morph into TikTok over 25 years. Imagine AI over the next decade.
Maybe it’s promise and limitations are best visualized in "The Infinite Conversation," an AI generated, endless conversation between German filmmaker Werner Herzog and philosopher Slavoj Žižek. The experiment shows stunning depth in AI voice construction, a convincing ability to grok the tone and substance of an imaginary discourse even if the conversation occasionally devolves into nonsense. It certainly points to a future where trust in sources is everything and where much of content we think of as valuable becomes commodity. I am imagining an Infinite Conversation between me and my mother….
50 new AI things
I wanted to start to think through where the impacts would be most immediate and profound, where new value would be created. I started to put together a list of stuff that has emerged recently.
Obviously, every creative and productivity tool, from Canva to Powerpoint to Google Docs to Final Cut will be redefined by AI helpers. Across every vertical software category, challengers will try to outmaneuver legacy tool makers by redefining tools around AI. Runway.AI is a good example in video editing. Algorithmic innovation and proprietary models will represent a new wave of value creation.
But the biggest change will be felt as organizations of every type optimize marketing and operational functions around AI. I believe the technology will force role and structural change much more profound than what was seen with the spread of personal computers in the workplace. The reason is pretty simple; AI will find applications in every department and job function but its potential to augment or replace human effort will run much deeper. Economics will reward organizations that put machines to work. Change will come fast.
Advantage will go to the learners and adapters, those that make and manage the models, people with the skills that AI can only inspire or optimize - the connectors, creators, makers, storytellers, inventors.
With that, a long list of new things that begin to reveal the future.
Creative tools
Facetune uses AI to improve selfies.
Don't like your own face. Find a fake one. Or make your own.
Luma Labs are smart Google and Apple people that use AI to turn 2D images into 3D manipulatable objects. Read about it here.
"Inpainting" and "outpainting" means using AI to magically retouch photos and video. ClipDrop does this and more including offering the ability to take a photo of any object and instantly turn it into a close cropped 3D model. Photoroom does the same.
Neural Radiance Fields or "Nerfs" use AI to take 2D things and turn them into 3D things.
Microsoft Designer starts with AI.
AR cut and paste.
Infinite Image from Runway.AI makes video an input to your imagination
DAIN uses AI to interpolate from low frame-per-second video into high FPS.
Automated marketing
Beautiful.AI is a pretty remarkable tool to make presentations look pro.
Showtime uses AI to make you a better presenter.
ShapeFactory and Looka use AI to make you a logo and more.
Copy Scouts is an AI blog post writer.
Use Astria to train the AI on your merch. Set your products in any environment.
HourOne creates talking heads from text.
EveryPixel searches all the databases of stock art better than Google Images.
CopySmith writes product descriptions for ecommerce sites. Copy.Ai does it too.
Automate talking heads with Synthesia
Durable will build you an entire website using AI. Or at least give you a starting point.
Clickable makes an ad in 10 seconds. Or at least gives you some ideas.
Automate talking heads with Synthesia.
CVitae will classify and format stacks of resumes, and track job candidates.
Personal assistants
Many people want to offer AI to help you write. Elephas, Jasper, Lex.
Upword creates personal summaries of long content.
Bearly.Ai puts all the Open.Ai writing tools at your fingertips.
OneSub is another attempt to use AI to gather and sift through news.
Rewind is a personal assistant for your Mac. Records everything you do. Compresses and stores locally.
Use AI to track your time without logging your time.
Build your social media bio with this.
Build your own assistant. This guy did.
Praktika is a personalized AI english teacher.
Lexica is a Stable Diffusion search engine that helps you find prompts.
And... stupid AI fashion advice.
Toys and more
AvatarAI shows what AI will do when we train it on image sets of you!
Replika is an AI text friend who won’t judge you.
A bunch of apps like Dream from Wombo and Picso show how diffusion engines are coming to simple consumer apps.
Infinite Monkey is an AI story generator.
MyHeritage has been around a while. It uses AI to animate your photos.
A guide to making your own diffusion engine.
An app that lets you run Stable Diffusion on your phone.
Click around a bit. It’s pretty cool.
Enjoy your weekend…/ Troy
(Note: I wrote more of the topic a couple of weeks back: AI and the Age of Creative Superhumans)