AI, Climate, and Energy (Resource Collection)

One of the most important things in the emerging AI-world is how we handle the ENERGY NEEDS of AI systems.

This actually invokes a much bigger question – how will we handle energy needs overall? How do we mitigate (and potentially reverse) the climate crisis? How do we build sustainability and resilience into our energy infrastructure?

Salon #6: “AI, Climate, and Energy”

This is a RESOURCE COLLECTION to accompany the Themesis AI Salon #6: “AI, Climate, and Energy,” held Sunday, May 12, at 2PM Eastern. (To receive an invitation, you MUST be Opted-In with Themesis; see the Opt-In form on our “About” page:

Meet Our Presenters

Our guest presenters are two people with extensive experience and knowledge in energy (especially green electronics – for Lee G.), batteries (especially EV batteries – for Brian T.), and a strong knowledge of how AI is being put to work today:

Lee Goldberg, “Full Spectrum Tech Journalist,” and
Brian Tieskotter, Staff Battery Engineer, Toyota Research Institute.

Brief Bios:

Lee Goldberg

Lee Goldberg is a self-identified “Recovering Engineer,” Maker/Hacker, Green-Tech Maven, Aviator, Gadfly, and Geek Dad. He spent the first 18 years of his career helping design microprocessors, embedded systems, renewable energy applications, and the occasional interplanetary spacecraft. After trading his ‘scope and soldering iron for a keyboard and a second career as a tech journalist, he’s spent the next two decades at several print and online engineering publications.

Lee’s current focus is power electronics, especially the technologies involved with energy efficiency, energy management, and renewable energy. This dovetails with his coverage of sustainable technologies and various environmental and social issues within the engineering community that he began in 1996. Lee also covers 3D printers, open-source hardware, and other Maker/Hacker technologies.

Brian Tieskotter

Brian is a Mechanical and Electrical Engineer who has spent his career focusing on product development and manufacturing of electronics, semiconductors, and Li-ion batteries. He has been especially interested in improving efficiency, reducing risk, and avoiding waste. Brian is currently a Staff Battery Engineer at Toyota Research Institute, where he explores better ways to manufacture and use batteries, as well as new ways to use machine learning and artificial intelligence in manufacturing.

Quick Get-Started Reading List (with a Few YouTubes to be Added Tonight)

Figure 1. Multiple energy sources will make it possible to create a robust and resilient energy infrastructure – where we can not only power our EVs from our home, but via a reverse-charging mechanism, draw energy from our EVs to support our homes or workplaces as a “backup” mechanism. (Figure courtesy Lee Goldberg; used with permission.)

This is the talk Lee gave on where batteries are headed:

Welcome to the EV (R)Evolution Sept. 14, 2023

Here is the article which the talk Is based on:

Advanced Battery Technologies Will Help Transform Our Energy Economy

From Brian T.

First, the Stanford University Human-centered AI Index report: https://aiindex.stanford.edu/report/

This is a basis for claims about AI cost, among other things.

AJM’s Note: This is a good find! This report presents ten distinct topics, and the coverage seems real/realistic. They note:

This year, we have broadened our scope to more extensively cover essential trends such as technical advancements in AI, public perceptions of the technology, and the geopolitical dynamics surrounding its development.”
Stanford University, Human-Centered AI: The AI Index Report: Measuring Trends in AI.

Second, an intriguing paper on a new architecture to replace MLPs (Multilayer Perceptrons) in some applications that is claimed to achieve better performance with much smaller size (but with higher compute cost than a MLP architecture of the same size): https://arxiv.org/pdf/2404.19756v2

AJM’s Note: This just came out last week; great find, Brian T.!

Liu, Ziming et al. 2024. “KAN: Kolmogorov–Arnold Networks.” arXiv:2404.19756v2 [cs.LG] 2 May 2024.

From their abstract:

Inspired by the Kolmogorov-Arnold representation theorem, we propose KolmogorovArnold Networks (KANs) as promising alternatives to Multi-Layer Perceptrons (MLPs). While MLPs have fixed activation functions on nodes (“neurons”), KANs have learnable activation functions on edges (“weights”). KANs have no linear weights at all – every weight parameter is replaced by a univariate function parametrized as a spline. We show that this seemingly simple change makes KANs outperform MLPs in terms of accuracy and interpretability. For accuracy, much smaller KANs can achieve comparable or better accuracy than much larger MLPs in data fitting and PDE solving. Theoretically and empirically, KANs possess faster neural scaling laws than MLPs. For interpretability, KANs can be intuitively visualized and can easily interact with human users. Through two examples in mathematics and physics, KANs are shown to be useful “collaborators” helping scientists (re)discover mathematical and physical laws. In summary, KANs are promising alternatives for MLPs, opening opportunities for further improving today’s deep learning models which rely heavily on MLPs.”
Liu et al. 2024. “KANs.”

AJM’s Note: This really IS a very different kind of architecture. Radically so. Will have to read/study. (This came in just before today’s Salon.)

Why EVs and Energy-Charging Stations Might Be More Important than We Think

News from Musk’s Tesla Corp. continues to go up-and-down. The latest (as of the last 24 hours) is that Musk is committed to superchargers. (Well, damn. He’d better be. Hard to run EVs without them.)

Here’s a useful and quick summary on the latest with Musk, Tesla Inc., and Superchargers:

Where things get REALLY interesting is the “reverse charging” model – the “vehicle-to-grid” charging option. SO many possibilities from this! A worldwide leader is Nuvve.

Fig. 2. Vehicle-to-grid charging options – where world-leader Nuvve allows greatest flexibility, including “selling back” your car’s energy to the grid. (Figure courtesy Lee Goldberg; used with permission.)

Sexier than You’d Think

So … nothing seems dryer, more banal, than … protocols.

But – protocols are ESSENTIAL for enabling different EVs to charge at charging stations OTHER than the ones created by their car’s manufacturer.

Case in point: charging your non-Tesla EV at a Tesla Supercharging Station.

You’ve got to match up the protocols.

Recent agreements will promote this.

One of the most important thing with Tesla Superchargers is that, over the past two years, they’ve been able to charge vehicles OTHER than Teslas.

Quick take:

Can any car use a Tesla charger?

Here is a 2-parter on the standardization of the Tesla charging interface – and its implications:

Using Your Car … to Charge Your House

Totally weird, isn’t it?

I live in Hawai’i.

My next-door neighbor, Jake, just put in a bigger and better set of solar panels on his house’s rooftop.

I look at the houses around me, and they ALL (including my own) have solar panels on the roof.

We charge at least our water heaters with solar; sometimes more. (So if it’s a cloudy wintery day in Hawai’i, the shower/bath is – at best – lukewarm. In summer, after a hot afternoon, the hot water is scalding.)

And my neighbor Jake is now talking about buying an EV next year … and plugging it into the house at night.

OK, so far … just common sense and nothing new.

But here in Hawai’i, we have a lot of people who embrace the “van life.” Who live “off-the-grid,” and that means … propane tanks for cooking, and probably cold wash water.

BUT … WHAT IF …

You could charge your household electricity from your car?
Your boss would PAY YOU to offer up your car’s electricity to offset the company’s electric bill during peak use hours?
Hospitals, emergency care facilities, and other disaster relief efforts could draw energy from cars — not just from generators.

And when needed, the cars could drive away, replenish, and return … essentially acting as mobile energy storage and transport devices.

Wouldn’t that be something?

And not totally outrageous.

Figure 3. Your car can become a part of a bi-directional energy flow system, offering energy as needed. (Figure courtesy Lee Goldberg; used with permission.)

Got five minutes? This YouTube hits the key points:

Vehicle-To-Grid: Could Your EV Power Your House?

Salon 6: “AI, Climate, and Energy”

This whole conflux of AI (with LLM’s mega-use of energy resources), climate (and the impacts of our energy use), and energy for EVERYTHING … is the subject of Salon 6, “AI, Climate, and Energy.”

To join, you MUST be a member of the Themesis community -meaning, signed up. Opted-In.

To do that, go to our “About” page.

Scroll down. Find the form. FILL IN the form.

DO the confirmation email, move the Themesis emails to your preferred Inbox, and … wait for it.

You’ll get an invitation.

Not free … but worth the price.

Intelligent conversation. A blessing and a privilege. All at once.