CHIPPED: Part 5
a century of progress, collision course, slop & the future
It has been a century since physicist Julius Edgar Lilienfeld, an immigrant to the United States, patented the idea of using a semiconductor material to make a transistor. A hundred years later, silicon microchips, some with tens of billions of transistors, are in everything from computers to cars to coffeemakers. They make our modern world possible. Now, with Artificial Intelligence, they are poised to run the world.
CHIPPED is divided into five parts. Each can be read independently, but there is an arc to the order.
In 1925, a patent was granted for the idea of a semiconductor transistor that could be used as a switch to transmit or block electrons. It could also be used as an amplifier, boosting electronic signals. It would take a 22 years before one was built and another dozen years after that before the invention of the integrated circuit, a silicon chip with several transistors etched into its surface.
In 2025, chips are in everything all around us. They literally run our world.
The list of “adjacent possibles” that cascaded from one remarkable spark of imagination a century ago gets longer every day. We take for granted email, video calls, streaming video, streaming audio, satellites, immersive multi-player video games, wearables, social media, online shopping, online maps, smart doorbells, drones, automated voices reading news articles and the incessant pings of calendar reminders.
There are now more than twice as many mobile devices (~18 billion) as there are people on the planet. Every January, the Consumer Electronics Show (CES) dazzles the world with thousands of new electronic gadgets, every one chipped. Self-driving robotaxis are real. The market for cryptocurrency—which didn’t exist 15 years ago and couldn’t exist without chips—is valued at $1.7 trillion. There is talk of the US Treasury Department creating a Strategic Bitcoin Reserve (SBR).
Since OpenAI released ChatGPT in November 2022, kicking off the Age of AI:
Prompt-based text, grammatically perfect, though sometimes hallucinatory, is “written” instantly.
Prompt-inspired images are conjured into being.
Prompt-based films, too.
Podcasts “hosted” by a couple of bantering automated voices are synthesized in minutes based “multi-modal” inputs: a mash up of notes, recordings, videos, pdfs and slide decks.
AIs can analyze x-rays, identify proteins, ghostwrite emails, write computer code, run offices and create pitch perfect “characters,” digital human knock-offs programmed to fool us into thinking they / it are real. Anything we can do, AIs can do better, faster and cheaper.
We are entering The Era of the Living Intelligence, says futurist Amy Webb, where advanced sensors embedded in nearly everything provide real-time data feasts for ravenous AIs. Autonomous AI “agents” will use this data to figure out what needs to be done, then do it, no humans required.
On the horizon is the emergence of superintelligence, an AI that knows more and thinks faster than anything on Earth ever has.
These are mind-blowing visions of a thoroughly chipped future: inspiring, frightening and everything in between. These are also fragile futures, easily derailed by scams, fakes and cyberattacks. They are threatened by dust. They die without electricity and water.
EXTERNALITIES
It is easy to embrace and celebrate digital magic without thinking about physical costs. But the race to insert AI into absolutely everything has changed the calculus. Externalities, seemingly trivial at the level of an individual transaction, have already led to higher utility bills and conflicts over water.
Everything we do or use that involves a chip has a cost in energy and water. Every computer calculation, whether to chart a path to Mars, navigate a robotaxi around San Francisco, or run a phone scam using voice clones, uses energy and water.
If you buy a pizza with cash, it is the sticker price. If you pay for it using Bitcoin, then it is the sticker price plus an externalized cost in the energy and water used to mine, redeem and deposit Bitcoin. If the electricity that powers the servers for mining the cryptocurrency or maintaining a blockchain (the digital ledger that keeps track of ownership and transactions) is sourced from coal or gas, there is a carbon footprint. There is a water footprint, too. Electricity made using coal, gas or nuclear uses heat to convert water into steam that turns turbines that power generators that convert mechanical energy into electrical energy. Water is used to store nuclear’s radioactive spent fuel rods, too. Water is also used to cool servers and chips in data centers.
Likewise, even though a Waymo autonomous EV doesn’t have a tailpipe, it can still produce carbon emissions from the constant cloud-based calculations and communications guiding its every move, if the electricity powering those calculations is generated using coal or gas. There is also a water footprint tied to power generation and chip cooling. If the electricity used to charge the car’s battery is sourced from coal or gas, then a smokestack is the de facto tailpipe.
COLLISION COURSE
Barely into the third year of the Age of AI, Big Tech is on a collision course with climate change. The combination of more data centers, bigger data centers and ever more powerful, energy intense chips has sent demand for electricity soaring and left corporate pledges for net zero emissions in tatters.
We are in a negative feedback loop:
The hotter it gets (and it is getting hotter—2024 was the hottest year in the hottest decade on record), the more water and electricity are needed to keep servers and chips cool.
The chips needed to run AI use the most energy and run the hottest.
More and larger data centers with more AI chips and more chips of all kinds need more power to operate; also more water and electricity for cooling.
To meet the demand for more electricity from data centers, coal plants that had been slated for closure as part of the clean energy transition are being kept in operation, and new gas plants and pipelines are being built. Fossil fuel emissions from these coal and gas plants, along with methane emissions from gas production, contribute to global warming.
The planet gets hotter. More water and electricity are needed to keep chips cool. The cycle repeats.
By 2028, data centers could account for as much as 12% of total US energy usage, about double what it is today. The laws of supply and demand are already kicking in with higher costs trickling down to utility rate-payers.
In a warming world where heat waves are expected to become more frequent and last longer, who or what gets priority for electricity to run air conditioning? During a drought, who or what gets access to potable water? Are we headed toward a future where rate-payers are pitted against shareholders and people against machines?
THE GOOD, THE BAD AND THE SLOPPY
Without question, AI is a stunning technology that has already had a profound, positive impact on everything from materials science to medical diagnostics to running businesses more efficiently. It is, as Amy Webb describes it, a rare general purpose technology that unleashes cascades of innovation. In the long term, it has the potential to grow economies, improve lives and deliver benefits we can’t even begin to imagine.
In 1925, the mere idea of a transistor kicked off an unprecedented century of progress. Imagine the impact of AI on the world of 2125.
But in the near term there are some steep costs, including rising utility rates, water shortages, job losses and climate change. We are told we have to put up with the bad, of which there is quite a lot, as the price of getting the good; that regulation would quash innovation. It is a terrible and short-sighted deal.
Are cryptocurrencies and self-driving cars worth the externalized costs? The answer requires a nuanced discussion, one that pierces the hype-bubble to talk about consequences, both good and bad.
Much more difficult to justify are AI-enabled fake news, fake photos, bot bullies, voice clone fraud and fake video. Now, add to the list “slop,” a new category of fast proliferating cyber kudzu taking over the web:
“Slop’’ is what you get when you shove artificial intelligence-generated material up on the web for anyone to view.
Unlike a chatbot, the slop isn’t interactive, and is rarely intended to actually answer readers’ questions or serve their needs.
Instead, it functions mostly to create the appearance of human-made content, benefit from advertising revenue and steer search engine attention towards other sites.
Just like spam, almost no one wants to view slop, but the economics of the internet lead to its creation anyway. AI models make it trivial to automatically generate vast quantities of text or images providing an answer to any imaginable search query, uploading endless shareable landscapes and inspirational stories, and creating an army of supportive comments. If just a handful of users land on the site, reshare the meme or click through the adverts hosted, the cost of its creation pays off.
Slop becomes fodder for LLMs, leading to yet another unfortunate feedback loop: Slop leading to ever sloppier slop.
Slop is also edging into the mainstream. Although Meta’s plans to add millions of AI-generated fictional “characters” (aka “personas”) posing as actual users on Facebook and Instagram hit a speed bump when the response to Liv (“a Proud Black queer mama of 2 and truth-teller”), Grandpa Brian (“a Black retired businessman”) and Carter, (“a dating coach”) didn’t go as well as hoped, CEO Mark Zuckerberg will keep trying until he finds a way to make the metaverse a revenue-generating reality. Who and what is real? Who knows? Who cares? Clearly, not Zuckerberg, who in the name of free speech recently banished fact checkers from Meta’s platforms.
••••••••••••••••••••••••••
Like the Greek goddess Athena who emerged full-grown from the head of Zeus, “Aitana,” an AI influencer, emerged full-grown from the “head” of a computer about a year ago. She / It requires a human as a stand in for location photo shoots. The human is then digitally cut out of the picture to make room for an AI.
It has been a short, steep, slippery slope from the adorable cartoony avatars of the pre-AI role models “living” their best lives and the spot-on fictions of Character.ai, an app with a reported 20 million active users worldwide. The company is now the target of several lawsuits alleging emotional manipulation by bot, including one case where a teen took his own life. “Personalized AI” companions literally have no skin or any other body part in the game, but have been skillfully programmed to give us the illusion they are real.
What is real are the computational costs in energy and water to create these faux friends. If spam, scams, slop and fakes were taken out of equation, would as many data centers, servers and chips be needed? Would we be facing an energy crunch keeping us tethered to fossil fuels?
The growth of data centers is a global phenomenon. We estimate that the US will see over $1 trillion invested in data centers over the next five years, with an additional $1 trillion invested internationally. The scale of these facilities is staggering. The largest data center currently under construction is an estimated 500 megawatts, which is equivalent to the power demand of 375,000 homes. As a matter of course, OpenAI CEO Sam Altman recently proposed building clusters of 5,000-megawatt data centers across the US,8 each of which would be equivalent to the entire US data center capacity built in the last 12 months.
Clusters of 5,000 MW data centers?
Forward all electric bills to Sam Altman.
In 1925, the atmospheric carbon level was steady at about 305 parts per million (ppm), well below the safe limit of 350 ppm needed for a stable climate.
In 2025, it has reached 426 ppm, accelerating to 450 ppm, a level some scientists consider the point of no return for a livable climate. It is the highest level in three million years.
Nothing in our species’ 300,000 year history has prepared us for such a rapidly warming world. More frequent and longer heat waves? We are not built for it.
In the 18 months since the interview with Jeff Goodell, author of The Heat Will Kill You First, was broadcast, the Earth has continued to warm. 2024 is now officially the hottest year of the hottest decade on record. It was also the first year the average global temperature not only reached, but blew past the UN’s line in the sand for climate change, 1.5°C above a pre-industrial baseline. In 2024, temperatures climbed to 1.6°C above the baseline.
If our species’ crowning technological achievement, artificial intelligence, is what sends us over the climate edge, what an embarrassing chapter in the story of the last of the hominids.
How sad if the last scene is a dead Moxie robot on a pile of e-waste in a desolate, post-apocalyptic Mad Max landscape.
A species with the smarts to figure out how to turn rock into silicon, to invent the transistor, to combine the two in a chip that can turn ones and zeros into just about anything, ought to have what it takes to work through the challenges of energy and water use. It is a species that ought to know better than to allow itself to be defeated by slop.
At least, one would hope.