For most of my life, the common lament was people weren’t engaged enough in politics. The hand wringers worried about the disengagement and apathy on the part of the electorate as a threat to the democratic process. Safe to say, that’s changed. In some ways, politics have eaten the world.
That’s why we are seeing a rush to define generative AI, much the same way as politicians pour money into defining a new opponent. John Fetterman’s campaign did this brilliantly with Dr. Oz, painting a well-known figure as an out-of-touch celebrity who actually lives in New Jersey and thinks an acceptable tailgate snack is a crudité.
AI has the hallmarks of the most significant advancement in the tech industry since the dawn of the iPhone and cloud computing, at least if you buy even half of what people like Microsoft CEO Satya Nadella peddle. The nascent tools have been dribbled out, notably Microsoft’s new AI-enabled Bing. The tech industry is no longer a bunch of naive kids in garages. They’ve been defined instead by opponents as anti-competitive surveillance capitalists with dangerous libertarian proclivities who are unable or unwilling to take into account the societal impact of their products.
The race to own AI is pitting the world’s largest, richest and most powerful companies against each other. I’ve watched enough seasons of Game of Thrones to know that no good happens when the powerful houses are warring, particularly for the villagers. Nadella has been admirably open about seeing an opening to make search far less profitable for Google – and wanting the world to know that it was Microsoft that “made them dance.” Google itself is uncharacteristically being defined as lumbering. It’s as if Google became IBM all of a sudden. Google’s scramble to demo Bard has been labeled by its own employees with the ultimate sin of being “un-Googley.” Perhaps more worrying, Google saw $150 billion erased from its valuation. Perception and vibes matter as much as free cash flow.
Politics is the pursuit of power. That’s fueling a lot of the early scramble to define these new AI tools as not wondrous advancements but ultimately as tools of misinformation, maybe some racism, and quite possibly profoundly dangerous. Microsoft’s Bing demo video has been deconstructed and found to be, at best, directionally accurate. If we cannot agree on a basic set of facts now, it’s hard to see how AI is going to make a consensus on reality more likely. More likely, we will see the developments of more tools to create a simulacrum of a preferred reality.
Of course, these models are flawed for the precise reason that they’re not actually sentient but are mimicking humans, who are themselves flawed and often wrong. These AI chatbots are using natural language to impersonate humans, giving confident and fluent responses. The early results are mixed, but then again, there are a lot of bullshitters out there, so it only stands to reason these models would have a fair bit of bullshit in them.
Lurking in the background is the fear of AI taking over. It probably doesn’t help that ChatGPT creator Sam Altman is also a prepper. What does he know exactly? Screenshots get bounced around of Bing’s chatbot getting combative. It calls to mind Microsoft’s disastrous early foray into AI chatbots in 2016, when Tay turned out to be racist. Reddit is filled with examples of hacking ChatGPT to go berserk. Cue Nightmare of LLM Street.
The first thing most reporters seem to do when getting their hands on these tools is to get them to deny the Holocaust. That’s sensible, but it also becomes political fodder for the techno utopians to dismiss the media as fighting a rearguard battle against tools that threaten their power – and in many cases existence. It feeds into the increasingly common belief among tech leaders that news publishers are out to get them and should be treated with extreme skepticism, if not outright hostility.
The results of search chatbots have been judged middling at best compared to the old approach. The Gartner Hype Cycle is now measured in days, not years. Coming on the heels of AI, there’s going to be an inevitable chorus of skepticism that this isn’t another tech financial bubble. Even worse, AI taps into our fears of being replaced by machines. We want machines to do the shitty work, not the stuff we find rewarding, as Juan Mendoza writes in The Martech Weekly:
By directly answering our questions, writing our stories for us, and creating images and videos, what it deprives us of is what makes us most human – the ability to create new things in the world. In a surprising twist of fate, AI’s purpose was always meant to eliminate human drudgery in an economy – transportation, updating spreadsheets, managing data, or customer service. But instead, the breakout-use cases are all focused on replacing how we create, the one thing that we’d want to do more of with our time in marketing.
I was a bit naive when I first saw AI chatbots and the potential. I thought this kind of tech would cut through the culture wars that have enveloped tech. Not so. The difficult questions should be asked of this technology and what safeguards are in place to keep them from running amok and making misinformation even worse. Ted Chiang, writing in The New Yorker, described ChatGPT using the metaphor of the compression of digital files, where the degraded fidelity becomes more perceptible the more the image is compressed.
It retains much of the information on the Web, in the same way that a jpeg retains much of the information of a higher-resolution image, but, if you’re looking for an exact sequence of bits, you won’t find it; all you will ever get is an approximation. But, because the approximation is presented in the form of grammatical text, which ChatGPT excels at creating, it’s usually acceptable. You’re still looking at a blurry jpeg, but the blurriness occurs in a way that doesn’t make the picture as a whole look less sharp.
As these AI chatbots are rolled out, there are many challenges, particularly around the current cost of processing queries using large language models. Publishers are rightly on alert for threats these days. (Experience helps.) Search has been the backbone of digital media, providing a steady stream of distribution to publishers in a mostly symbiotic relationship with Google. The change of the search interface to accommodate AI chat will make that less effective. Some will click on the citations, but many will find the answer good enough. Like much of life, search is a real estate game. The choicest real estate will become curtailed. The bottom half of the first page of results becomes the second page and so on. Should these tools take off, SEO will become more challenging. Nothing is permanent.
Going solo
On this week’s episode of The Rebooting Show, I spoke to Darren Samuelsohn, a longtime politician reporter who was most recently head of Insider’s DC bureau. Darren and I spoke about his new newsletter devoted to journalism, the decision to take an independent path, and the topsy-turvy career that’s journalism. Listen on Apple or Spotify.
House of Kaizen is an extension of your subscription revenue growth team. Decades of subscription growth experience are brought to bear to get you where you want to be faster with more confidence and efficiency. Through its SubscriptionWorks community, House of Kaizen brings together subscription publishing professionals to learn from each other. Over the coming months, SubscriptionWorks is hosting Q&As with subscriptions experts at The Washington Post, AARP, One Medical and more. Learn more here.
Recommendations
In retrospect, maybe the video conferencing tool shouldn’t have been judged more valuable than the largest energy producer in the world. Such were the pandemic delusions. Benedict Evans has published his annual tour de force of macro trends in the tech industry. The big theme for tech and media is the end of free money forcing a return to normal trend lines. “We were gamblers at a craps table crediting our resilience and daring, while the casino was pumping pure oxygen through the A/C vents,” writes Scott Galloway. .It turns out the pandemic didn’t accelerate trends as much as many hoped. What we’re seeing now with layoffs and trimmed initiatives is the outgrowth of that.
Early in the pandemic, the Financial Times painted a grim picture of what lay ahead, based on the experience of the Great Influenza, with the note that while millions might die, entertainment would explode. That turned out prescient, and coincided with boom in streaming services, which are now beating a hasty retreat to focus on profitability over growth. This “180” even includes the unlikely turn back to theatrical releases. Even the video games industry is scaling back to adopt less risky strategies that emphasize tried-and-true franchises. Disney, with Bob Iger back in control, is cutting 7,000 positions as it streamlines its operations.
When doing cuts, many companies are adopting Mark Zuckerberg’s mantra of a “year of efficiency.” That translates into targeting middle managers (and reeling in DEI departments).. Large companies are filled with managers who manage managers. Meta is asking managers to start making things. Elon Musk’s deep cuts at Twitter are an inspiration to embrace “the conviction that the world’s top tech companies need little more than core engineering teams.” The downside: Research shows these kinds of short-term cost savings moves hurt in the long run.
The good news for the middle managers is this is a very long game. The 60-year career is apparently here, so there’s still plenty of time to catch up. Great news. But it does point to something of a shift: careers are not linear. The “career path” was always a bit of a misnomer, unless that path was incredibly winding, with dead ends and periods where it is overrun with weeds and stumps. “Instead of advancing vertically up a single path, for instance, people will need to move sideways—and even down at times—as they traverse different jobs and multiple careers.”
Thanks for reading. Send me a note with your feedback: bmorrissey@therebooting.com. You can also just hit reply to the email.
If you’re looking to reach 14,000 publishing professionals, you should consider sponsoring The Rebooting to showcase your expertise in finding paths forward for building sustainable media businesses. Check out the sales kit and get in touch.
AI is actually better understood when we describe it as a machine learning, allowing us to further explain the different types of machine learning like reinforced learning. I also think phrases like 'synthetic' are helpful since that gives immediate meaning to the types of responses we generate from some 'AI' tools. Anyway, good article.