Welcome to Stuff I Thought About Last Week, a personal collection of topics on tech, innovation, science, the digital economic transition, the finance industry, and whatever else made me think last week.

Click HERE to SIGN UP for SITALWeek’s Sunday Email.

In today’s post: automation is being overlooked by many businesses clinging to people-intensive processes instead of reimagining and freeing employees for higher value work; the design community is iterating quickly with open tools for AI-assisted creation, forcing philosophical questions about art and challenging the old way of doing things; 3rd-party data brokers continue to threaten privacy and safety; Google's Wing still sees drones as the key to last-mile delivery for most packages; when a chicken talks to a cow; and, much more below...

Stuff about Innovation and Technology
BotRistas
The quit rate for new baristas at Starbucks has risen from 10% to 25%. One of the reasons behind the higher attrition is the economy-wide job hopping for higher wages at a time when the labor force is structurally shrinking (the labor force is still rebounding from COVID, but the long-term trend remains problematic). Another factor in employee churn is the increased strain from higher throughput digital ordering, which has also notably facilitated complexified customizations. This WSJ article highlights how founder and interim CEO Howard Schultz is working to make the barista’s job easier with better tools and store layouts. While those process tweaks might pay off, they seem to miss the point: Starbucks has a highly automatable product – the goal shouldn’t be to make baristas more efficient, but to replace them with technology. I suspect we will see this pattern play out at more and more legacy businesses, with long-time management teams wedded to a romantic, people-based version of their business instead of facing the reality that automation, robotics, and software need to supplant workers performing repetitive, mundane tasks. Simply equipping baristas with more aerodynamic buggy whips (so to speak), will not be a long-term solution for handling shrinking labor, rising costs, and digitally increased business/throughput. Starbucks also faces the challenge of unionization efforts among its employees. Unions are making a fairly big gambit because the more they press, the more they should expect to see automation eliminate the jobs they are fighting for. Across the economy as a whole, there’s clear potential for a classic power law to develop. At the head will be businesses operating at large scale that successfully implement cost-effective automation. Then, you will have a long tail of smaller businesses still capable of providing a personal experience – and charging a premium to cover the labor costs. Starbucks used to be able to provide a more personal experience at scale, but current challenges seem to be forcing an existential technological shift. I suspect Starbuck's apparent stuck-in-the-middle strategy of trying to get the best of both worlds will be an interim solution, at best.

AI Design
Last week, I talked about the rapid iteration of AI tools for the design world, noting a potential inversion in how we create new ideas:
In the past, we might have sat down with a sophisticated design program, sketched out a theoretical wind turbine or heat exchanger, and then simulated how it might function in the real world. But, with machine learning and AI, we can instead say something more akin to: here is what the world looks like, now go and create the best solution. It effectively inverts the job of design from “I have an idea” to “what should my idea be?”
I came across a few examples this week of some impressive design tools (e.g., Twitter posts here and here) that show how quickly AI models and design are coming together. It will be interesting to see how the traditional tool vendors like Adobe choose to either embrace or resist this movement to automate and open source design tools. Certainly, Adobe’s software will be used for years to come, and they can incorporate many of these systems and models, but can they be an open platform for innovation of the tools themselves? That’s typically a hard shift for companies to make, e.g., how traditional software companies rallied against open-source operating systems and apps. The examples I see of the new design tools feel like foundational layers of technology for AR and VR, which will require the ability to iterate and render overlays on the fly. Related, this article has a series of fantastic side-by-side renders comparing the different design aesthetics of Midjourney, DALL-E, and Stable Fusion (a newly released open-source image generator on which you can build apps).

AI Art
Recently, an AI-generated submission won first place in the digital category at the Colorado State Fair and received much criticism for not being “real” art. For tens of thousands of years, the tools used by artists have evolved (probably for longer than that, but the earliest cave art we’ve found is estimated at ~60,000 years old). With the rise of transformer AI models like Midjourney, which was used to win the prize mentioned above, we are (yet again) faced with philosophical questions over the relationship between creator and art. If moving from charcoal chalk to paint and canvas made it easier for an artist to render their vision in the real world for others to enjoy, was that cheating? Most of us would say no. But if an AI model creates an image or movie from simple human prompts by leveraging prior artworks from others, is that cheating? (There seems to be a gray area between taking inspiration from and directly copying/manipulating prior art). I argued a few weeks ago that these tools actually give us more agency over our creative endeavors, meaning they might be able to better translate what is in our heads and hearts into a medium that others can also connect with and enjoy. This appears to be the case with the creator of the prize-winning Midjourney creation noted above, as he spent 80 hours with 900 iterations before arriving at the vision in his head. In general, true art is really hard to achieve. The end product might make it look easy, but significant work goes into perfecting the creation of quality works. A point I’ve made before, referencing Zen and the Art of Motorcycle Maintenance, is that anything can become Art (i.e., Quality, per Pirsig) if you care enough about it. You can use AI to generate images (and soon videos) relatively easily, but generating what we currently consider to be Quality art is likely to be a far longer and involved (and sometimes tormenting) process. And, as noted above, these new AI models (and the video tools built on top of them) seem like they are the basis of the tools that will build AR and VR worlds. Of course, it’s not just visual arts that will face these questions. Composers, architects, engineers, and any other creative endeavor will be aided or completely replaced by AI. Could an AI composer divine the fractal patterns in classical music and give Mozart a run for his money? In an example that blurs the lines between artist and AI, the deepfake company Metaphysic has been competing in America's Got Talent this season with singers performing in real time as other people. As I mentioned last week, AI assistance also creates complex issues of prior art, copyright, trademarks, patents, etc. because, in some cases, we won’t even be able to trace what went into a new song, movie, design, etc. Perhaps nothing has a larger influence on the future than art because nothing touches our hearts or moves us more than getting that glimpse into someone’s creative mind. Artists are also frequently at the edge of culture, pushing the limits. I am not sure how we will judge AI-designed creations in the future, but I think by examining the trends around them we have a good shot at seeing where the future might lead us. For now, I’ll take my direction from this Kurt Vonnegut quote that I put forth often: “Practicing an art, no matter how well or badly, is a way to make your soul grow, for heaven's sake. Sing in the shower. Dance to the radio. Tell stories. Write a poem to a friend, even a lousy poem. Do it as well as you possibly can. You will get an enormous reward. You will have created something.”

Precise, Near-Real-Time Location Data for Sale
EFF exposes Fog Data Science, a company that buys data from many of your apps through data brokers, tracks you in real time, and then sells it to law enforcement (no search warrant required) or any private investigation firm. The “pattern of life” data, which include latitude, longitude, time, and device ID going back to 2017, is searchable in an online portal for clients by device or general area. Subscribers can query the database 100 times per month for a cost of $6,000-9,000 per year. Fog is believed to be an affiliate of Venntel, the largest supplier of location data to the government. If you care about privacy, it’s important to make sure all 3rd-party apps you have only use location data while running, and you should take advantage of new controls like Apple’s App Tracking Transparency opt-out. It’s likely most of Fog’s data comes from ad trackers utilized by the complex data broker and ad platform ecosystem that, as I’ve said before, should be regulated out of business (EFF mentions that the Starbucks and Google’s Waze map apps could be two of the sources of data used). In the meantime, ongoing changes on iOS and Android will hopefully go a long way toward eliminating the practice. I’m tempted to think there is an inevitability toward increased privacy, but the last two decades seem entirely contradictory.

Diversifying Delivery Drones
Google’s Wing drone delivery business thinks that a proliferation of different drone designs will be the key to allow drones to deliver the vast majority of our ecommerce shopping more cost effectively than cars or trucks. The service continues to operate in Dallas with Walgreens and is hoping to slowly expand to more locations.

Miscellaneous Stuff
DeepSqueak
Back in #278, we mentioned researchers using advanced algorithms in an attempt to create a communication tool with sperm whales. Scientists are now using machine learning to analyze large troves of data on animal communication in what might lead to language models, and (who knows!) maybe even the ability to communicate with – or at least understand – what various species are saying. This NYT article looks at naked mole rats, noting that researchers have discovered that different colonies each have their own dialects that are culturally transmitted from generation to generation. One program, called DeepSqueak, can “automatically detect, analyze and categorize the ultrasonic vocalizations of rodents.” Machine learning algorithms applied to Egyptian fruit bats “could distinguish, with 61 percent accuracy, between aggressive calls made in four different contexts, determining whether a particular call had been emitted during a fight related to food, mating, perching position or sleep.” Imagine the future universal animal-to-animal translator that will let a chicken finally tell a cow what’s on its mind. The TikToks will be endless.

The Dudes Are Not Yet Abiding
A Gallup poll revealed that more Americans reported smoking marijuana (16%) than tobacco (11%) in the prior week while 14% of respondents consumed edible marijuana. I’m not sure how accurate these numbers are, but there sure seem to be a lot of slow drivers on the road whenever I leave the house, and not all of them are live streaming on TikTok. Wasn't there a hippie dream that everyone would chill out, and we’d have world peace from marijuana usage? Maybe there’s some chill tipping point we have yet to hit, but the world sure feels like the opposite of the prophesied nirvana. The title of the Gallup report is: “Americans Not Convinced Marijuana Benefits Society”. While those polled were about equally divided on the question, the numbers skewed quite strongly one way or the other based on whether or not the responder used marijuana (I’ll let you guess which way).

Stuff about Geopolitics, Economics, and the Finance Industry
Cloud's Water Footprint
NPR reports that 20% of data centers in the US operate in regions with moderate to high stress watersheds, which is concerning given that a "mid-sized data center consumes around 300,0000 gallons of water a day, or about as much as 1,000 U.S. households." Globally, drought and ongoing power production issues may cause the Internet to be rerouted to regions with better climate stability, to the extent that they exist. We will likely see rules about data sovereignty (requirements for storing data in the user's country of residence) relaxed in order to keep the cloud running.

Tech Leaving China
The NYT reports that tech companies are slowly moving production out of China due to ongoing pandemic challenges and rising geopolitical tensions. India and Vietnam are two of the largest beneficiaries for now. China continues to effectively withdraw itself from the global economy, increasing the range of outcomes across many fronts.

✌️-Brad

Disclaimers:

The content of this newsletter is my personal opinion as of the date published and is subject to change without notice and may not reflect the opinion of NZS Capital, LLC.  This newsletter is an informal gathering of topics I’ve recently read and thought about. I will sometimes state things in the newsletter that contradict my own views in order to provoke debate. Often I try to make jokes, and they aren’t very funny – sorry. 

I may include links to third-party websites as a convenience, and the inclusion of such links does not imply any endorsement, approval, investigation, verification or monitoring by NZS Capital, LLC. If you choose to visit the linked sites, you do so at your own risk, and you will be subject to such sites' terms of use and privacy policies, over which NZS Capital, LLC has no control. In no event will NZS Capital, LLC be responsible for any information or content within the linked sites or your use of the linked sites.

Nothing in this newsletter should be construed as investment advice. The information contained herein is only as current as of the date indicated and may be superseded by subsequent market events or for other reasons. There is no guarantee that the information supplied is accurate, complete, or timely. Past performance is not a guarantee of future results. 

Investing involves risk, including the possible loss of principal and fluctuation of value. Nothing contained in this newsletter is an offer to sell or solicit any investment services or securities. Initial Public Offerings (IPOs) are highly speculative investments and may be subject to lower liquidity and greater volatility. Special risks associated with IPOs include limited operating history, unseasoned trading, high turnover and non-repeatable performance.