Bluesky flies high, Gemini digs deep, and Sam Altman’s $7 trillion poker face
Is Sam Altman bluffing?
Everything I know about poker I’ve learned from movies and TV shows. (This is also true for time travel, international drug smuggling, and Japan.) In brief: poker is a card game in which you win by adopting a Lady Gaga-style “poker face” and having the confidence to wager a giant pile of gambling chips even though your hand of cards is weak (I believe this is called “bluffing”) in the hopes of frightening the other players into quitting because they must match your bet. There are probably some rules about the cards themselves, but those never seem to matter.
Whenever I watch a fictional poker scene, the camera lingers on the cards being held as though this should be meaningful to me. But it isn’t. Are aces high or low? No idea. But what I do know is that the player with the deepest pockets usually wins.
Why am I talking about poker? Because on February 8, Sam Altman, who is CEO of OpenAI, the maker of ChatGPT, reportedly talked to international investors about the need to raise up to $7 trillion (that’s US dollars – Canadian dollars don’t even go that high) to boost the world’s chip building capacity. This investment is apparently required to train and power future iterations of LLMs as well as other machine-learning platforms and products.
Let’s put this chunk of cash in perspective. Global chip sales are projected to reach $1 trillion per year by 2030, so this is seven times that amount. I sneakily went behind Sam Altman’s back and asked ChatGPT for some comparisons. Then I asked Google’s Gemini and the answers were more informative. Or possibly made up. Here’s what the LLMs told me:
$7 trillion is about one-third the output of the US economy and half of the European Union’s GDP. $7 trillion is around double the amount of debt relief provided by the World Bank and IMF to low-income countries. It’s more than the 2022 GDP of all 54 African nations combined. You could provide basic necessities like food and shelter for everyone living in extreme poverty globally for over 10 years. You could invest in clean energy and combat climate change significantly (the International Energy Agency estimates that $1.7 trillion needs to be invested annually in clean energy to achieve net-zero emissions by 2050).
So $7 trillion is a lot. And here’s where we return to poker. Sam Altman is pushing the world’s biggest pile of chips (pun unavoidable) into the middle of the table and hoping that the other players will fold their cards. After all, OpenAI is already backed by Microsoft, the world’s most valuable company. Maybe this surreal sum of money is an attempt to bluff the other gamblers at the AI table into thinking that his hand is already unbeatable.
Of course, the main question we should all be asking is: why not invest $7 trillion in people? A chatbot told me that we could fund the entire healthcare systems of the top 10 richest countries in the world for a year with that amount of money. What about the poorest? Maybe people in the global south wouldn’t risk their lives as economic migrants to the north if their health and education needs were provided for? But no, let’s see how much we can invest in artificial people.
I totally forgot my other source of poker information: the 1978 hit single The Gambler by country music legend Kenny Rogers. To sum up: You’ve gotta know when to hold ’em, know when to fold ’em, know when to walk away, and know when to run. So when one man pushes $7 trillion worth of chips into the middle of the table, should we hold, fold, walk away, or… maybe run like hell?
Bluesky opens
You might have already heard of Bluesky but, if not, I’m here to part the clouds. Twitter’s first CEO, Jack Dorsey, left the company in November 2021 and was a key player in the official launch of the Twitter/X spinoff BlueSky that same fall. Having witnessed the issues faced by Twitter even before it became a monster/failure under Elon Musk, Dorsey saw the need for a truly decentralized microblogging social network. Set up as a public benefit corporation (akin to for-profit socially progressive companies such as Patagonia and Ben & Jerry's), Bluesky subscription was by invitation only until earlier this month.
Although the company hasn’t released any audience numbers since 800,000 new users registered on the first day of general availability, Bluesky appears to have a fighting chance of someday challenging Twitter/X and Threads. As a Twitter user since the pre-troll days of 2007, I can confirm that Bluesky replicates the “pure” social sharing experience that we were promised. But beyond that, Bluesky’s principal advantage is that it is an open network (like email) meaning that users are not stuck inside a walled garden. The company commissioned a cute comic strip to explain the benefits. If you want to check it out, I’m on Bluesky here. Stop by and say hi!
Reddit Makes Bank
For many people, Reddit flies under the radar. It’s not a social network like Facebook. It’s not a content platform like TikTok. And it’s not an image or video-sharing site like Instagram and YouTube. It’s basically an online forum, which was a commonplace beast in the Wild Wild Web of the late 1990s and early 2000s. But guess what? Reddit is one of the top dozen or so most-visited non-porn websites in the world. And as its owners seek an IPO valuing the company north of $5 billion, Reddit has reportedly signed a deal with Google worth about $60 million per year may boost that valuation.
LLM training data sets are becoming subject to copyright claims because most AI companies have not sought permission to use online content, so the Reddit deal may signal a sea change in the way that machine learning will be fueled in the not-too-distant future. My only question is: should Reddit have asked for more? OpenAI is now apparently worth $80 billion, and if that number is too big for you to get a handle on, it is well over a thousand times more than $60 million. Whether or not Reddit’s deal turns out to be a good one, let’s not forget that the people who will get zero dollars for their work are the writers* of the Reddit content that is being sold.
*Post-publication, it was revealed that the top 75,000 Redditors will be able to purchase shares during the company’s upcoming IPO.
Regular Folk AI
As I explained in a recent LinkedIn post, Google’s Bard chatbot has been rebranded as Gemini. I often cover AI from a negative perspective in this newsletter (usually because it’s overhyped) but I don’t mind giving credit when a new tool appears that genuinely enhances productivity without a human cost. Over the last week and a half, whenever I’ve compared Gemini’s output to ChatGPT or Microsoft’s Copilot for the same prompt, Gemini is better in every respect. It organizes its response in a way that provides more useful information than you were expecting. Basically, it goes beyond the prompt. To be clear, I’m comparing the free versions of these platforms. Another thing that’s cool about Gemini is the little G button at the bottom of each response which automatically googles the information when you press it and then clearly indicates which parts of the response are based on real sources and which are a bit iffy. This is extremely helpful because all chatbots make things up. (It’s not their fault – they know nothing about the world.)
These features are already impressive, but now you can integrate Gemini with the full suite of Google products, such as Gmail, Docs, and YouTube. This opens up a whole new dimension of everyday usefulness for regular folk. Prime example: you can ask it to summarize recent emails from a specific person or client and it will scan your Gmail and output a short list. This should save Gmail users tons of time searching through emails for that missing piece of info, especially if they have forgotten which person in the client’s organization sent it (which never happens to me, oh no siree). To check out how it works for yourself, all you need to do is enable extensions in the gear icon on the Gemini homepage.
Charlie Harding, songwriter and podcaster
One of my favorite podcasts is Switched on Pop from the Vox Media Network, New York Magazine, and Vulture. Co-hosted by musicologist Nate Sloan and songwriter Charlie Harding, the podcast is “about the making and meaning of popular music.” Each episode digs deep into pop hits past and present, balancing cultural context with musical insight. Obviously, I was absolutely delighted when Charlie agreed to answer a question for Discomfort Zone.
Q. Songwriters have reused chord progressions and riffs since pop, rock, and blues began. Sampling and interpolation have become commonplace over the last three decades. Do you see the use of AI to write songs as the next phase in an ongoing trend toward creation through repurposing rather than originality?
A. The goal of popular music is simple: reach the largest audience with the strongest message. There’s no magic formula to getting this right. If there were, I’d be using it right now, and I'd be enjoying my mansion in Malibu. But if I had to distill the essence of a hit song, it’s about creating the perfect balance of repetition and variation.
Throughout time immemorial, musicians have used and recycled genre clichés, giving audiences something familiar to latch onto. Whole eras of music have been based on simple musical formulas, whether it’s the I vi IV V chord progression — also called "ice cream changes" — that made '50s R&B so sweet, the Amen break forming the basis of so much electronic and hip-hop music, or the requisite instrumentation to build a rock band: guitar, bass, and drums. Pop music is fundamentally built on reusing old ideas.
Now, with the assistance of artificial intelligence, working musicians and aspiring artists may be able to leapfrog the first steps of the creative process, using AI to create the formulaic bits of a song, so the artist can focus on the variations that communicate some emotional truth. There’s a long history of technology replacing musicians and increasing the speed of production. But these tools — drum machines, autotune, and digital audio workstations — have also fueled new modes of creative expression. Perhaps, in the coming months, as AI accelerates in usefulness, it will displace more than we can reasonably foresee. But I’m an optimist about the future of music.
If there is a proliferation of sound-alike songs, audiences will tire of this repetition and seek more musical variation in songs that communicate stronger human emotions in ways relevant to the times. Most importantly, music isn’t a solitary pursuit to be created by AI and consumed alone. Music is a site of participation, collaboration, dance, and celebration, the elements that make us most human.
Charlie Harding is co-host of the podcast Switched on Pop distributed by New York Magazine, co-author of Switched On Pop: How Popular Music Works and Why it Matters winner of the 2021 PROSE Award in Music & the Performing Arts, and an adjunct professor at NYU’s Steinhardt's Department of Music and Performing Arts Professions.
Mixed bAIg
In this section, I usually write about a single piece of news or development that is shaking up our culture, but honestly, there have been so many dystopian scenarios involving artificial intelligence over the last few weeks that I decided to compile them and allow the full force of discomfort to wash over you. (Sorry.)
Worried about losing your job to AI? Lucky you aren’t a skilled forger! Neural networks such as this one can produce a fake ID in seconds for only $15, allowing you to launder money, buy illicit goods, and whatever else sketchy people do.
Poets are being hired by Silicon Valley companies to help train LLMs. Job postings have been seeking applicants who write fiction, not only in English, but in Hindi and Japanese, to feed the machine-learning beast. The upside? Poets get paid. The downside? Poetry gets shitty.
It doesn’t get much more dystopian than this video (which is definitely real) of autonomous humanoid robots going about their silent business. Norwegian-Californian tech company 1X (no relation to Elon) “designs androids that work alongside people, to meet the world’s labor demands and build an abundant society.” That work includes baking cookies.
Finally, an antidote to data-driven discomfort: a recent anti-racism campaign in Barcelona used the power of handwritten signage to personalize immigrant-run businesses in the city. The reception has been so positive that it has spread to other parts of Spain.
When and How Should Corporations Speak Out?
Alison Taylor, associate professor at New York University, is the author of the Harvard Business Review book Higher Ground: How Business Can Do the Right Thing in a Turbulent World. In this episode of the HBR IdeaCast podcast, she speaks to host Alison Beard about the minefield that companies and other public-facing organizations must now walk (or run) through when they decide to comment or stay silent on political issues such as conflicts in Ukraine/Russia and Israel/Palestine, Black Lives Matter, climate change, transgender rights, and a host of other contentious matters.
Social media and employee activism now mean that the days of a quick PR campaign are long gone. Taylor explains why staying silent is often not an option and offers strategies that organizations can use to align their policies and practices while hopefully avoiding a reputational nightmare
Head in the Sand?
Exceptionally, the next issue of Discomfort Zone will drop in three weeks, not the usual two. Why? Spring Break, baby!
To keep you chuckling in the meantime, here’s the latest Midjourney mishap featuring a man, an ostrich, a head, and some sand, but not in the configuration that the user had hoped.
Send me your comments on Discomfort Zone at john@johnbdutton.com and please connect with me on LinkedIn if you haven’t already.