SITALWeek #411
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: batteries had a big smoothing impact during 2023's record heatwave; AI is slowly gaining its senses, traveling a path toward humanity; the shift from multitouch to chat-based interfaces may open up the phone market to disruption; cat states; and cloud KYC.
Stuff about Innovation and Technology
Electric Pinch Hitters
Grid-scale batteries are starting to have a meaningful impact on power consumption, providing a boost at sundown when solar ebbs yet energy demand for air conditioning across the US is still high. So far, as the WSJ reports, the grid has been able to avoid strain-related blackouts despite the warmest summer on record. For example, batteries accounted for 5.2GW, or ~18%, of energy supply at sunset in California last Sunday (data from CAISO, you can select the date from the drop-down menu).
Vision Comes to LLMs
AI will slowly gain the sensory input that humans and other creatures enjoy (see AI Awareness), and will eventually gain many superhuman senses (see How to Interact with an LLM). An important milestone on this path was achieved last week as the most popular LLM, ChatGPT, announced several features for its paid users, including the ability to see (via photo access), listen, speak, and render images. Some of these features were already available in BingChat, but the combination in GPT4 is notably more powerful. OpenAI also reintroduced web search to GPT4; however, in its brief absence, it appears many sites have altered their robots.txt instructions to forbid scraping. Not everything is integrated yet; for example, GPT4 on a mobile device offers voice, but not web search, and GPT4 in a browser has web, but no voice. There are five voice choices for GPT4, and I don’t want to say that “Sky” sounds like Samantha from Her, but the resemblance between the two is not zero. One thing that’s obvious from using the new features is that we very much need to give chatbots real-time vision, which could be accomplished with glasses that have cameras, or, per Spike Jonze’ solution, shirts with chest pockets (or, maybe phone lanyards will be a thing…). This video from OpenAI shows how the new integrations can help you answer real-world questions that require vision. It’s beginning to get quite easy to imagine an ever-present human-like assistant, particularly once OpenAI turns on the long-term memory feature.
LLM Friends
AR gaming and mapping company Niantic posted a whimsical take on AI assistants in the video embedded in this blog post. This example illustrates a concept I’ve been thinking about with respect to LLMs: why have one assistant when you can have dozens of LLM “friends” that fulfill different roles for various circumstances – e.g., a local expert, a therapist, a doctor, etc. Imagine if all of the voices in your head were actually useful! Meta officially announced their rumored chatbots that I highlighted in Will She Be a Llama?. While the bots are text based for now (and include celebrity friends like Snoop Dogg), Meta says to expect voice soon. LLM friends would effectively personify the Internet, i.e., these personality-based chatbots would become new interfaces for search, social, apps, and other common Internet use cases.
AiPhone
The phone market is wide open for disruption as we move from multitouch to a chat-based AI interface. Google is well positioned with its LLM efforts, but Apple could be vulnerable. The phone itself needs to become more aware 24/7 with vision and audio, and it will need to interface with AR glasses and provide more onboard processing. I highlighted this potential disruption in Discovery Engines back in April, and, last week, news broke that OpenAI’s Sam Altman and former iPhone designer Jony Ive are potentially working on an “iPhone of artificial intelligence”. Status quo will be hard to disrupt; but, for the first time in well over a decade, there is a major UI transition opening a path for disruption.
Tax Evaders Beware
The IRS is opening audits into 75 of the largest private partnerships (over $10B in assets) and will send 500 compliance alerts to other partnerships, thanks to its new AI tools. Previously, partnerships were too large and complex – and the IRS was too short staffed – to find anomalies. IRS Commissioner Daniel Werfel says: “modernizing the I.R.S. is good for everybody.” As trillions of new LLM-based “workers” enter the workforce, just think of all the great things they could accomplish – like lawsuits and audits! But, seriously, this is just the start. There will be a major expansion of the overall economy as this marketplace of bots takes off, effectively creating an entirely new digital economy many times larger than our analog one. See Simulacrum for a deeper discussion on this topic.
Miscellaneous Stuff
Cat States
Errors are the enemy of quantum computing. Essentially, the current problem with quantum computing is that the number of traditional (non-quantum) processors needed to offset the errors of quantum processors multiplies so quickly that quantum computing is a non-starter. A French startup called Alice & Bob (I first met the textbook characters Alice and Bob while getting my degree in astrophysics, but I have not run into them in decades) has come up with a way to encode a qubit so that its two states are extremely separated from each other. This setup contrasts to how qubits are typically encoded with more proximate variations (e.g., energy levels of a molecule) that are prone to erroneous bit flip. Alice & Bob use so-called “cat states”, in reference to Schrödinger’s cat, which is either dead or alive – two extreme situations. Cat states appear to be quite effective at reducing bit flip, thus improving data integrity. Unfortunately, these tricks are a drop in the ocean when quantum computers need to incorporate several orders of magnitude more error-free qubits to be viable.
Stuff About Demographics, the Economy, and Investing
Cloud Control
A year ago, I suggested in Know Your Cloud Customer that the US was making a mistake by focusing on advanced semiconductor sanctions while allowing access to US-based cloud computing for anyone around the world. The White House is finally considering KYC for cloud computing customers. In the meantime, the US is actively selling the most advanced AI to enemies of the West, no questions asked.
✌️-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.