Discover more from hi, tech.
No. 158: 🤖📝 Buzzfeed AI - "A Race to the Bottom"
+ Twitch Ads, Yandex leaks, The US vs. Google, Amazon pills
And welcome to hi, tech. 158. Yes, I was watching the Australian Open earlier.
We’ll be looking at:
🤖 Buzzfeed cut its staff by 12%, then announced it would use AI to create “personalised quizzes”. Their stock is booming.
So, what should we learn?
But realistically, what will we choose to learn instead?
👨⚖️ The US government is targeting Google’s ad business. It could win.
🔥 Leaked documents reveal Yandex’s ranking factors.
💊 Amazon moves into prescription subscriptions.
🎮 Twitch is revamping its advertising options.
🖼 Shutterstock embraces generative AI.
🐻🤳A selfie-taking bear.
If you’re not a full subscriber to hi, tech., you missed two sensational pieces this week.
First, we looked at how OpenAI will revolutionise Microsoft Azure.
Then, we predicted five ways Google will strike back.
If you are interested in keeping up with the latest tech analysis, the January special offer is still live!
You can get 20% off forever here 🤓👇🤓👇🤓
And we’re doing group discounts when multiple people sign up from the same organisation:
If you’re not ready for a full subscription but you can appreciate that this publication is both valuable and the result of mucho, mucho work, you can buy me a coffee here:
Take a little look at Buzzfeed’s stock price this week:
Yep, it’s up 325%. Two factors combined to create this growth, with the latter being the dominant force:
Buzzfeed, which has been struggling for a while, cut 12% of its staff in December
This week, Buzzfeed’s CEO announced to the company that generative AI would move “beyond R&D” to start creating content this year
All of the tech companies that laid off staff recently have seen a short term boost to their stock. For Buzzfeed, the impact was delayed and dependent on the announcement of a new, AI-driven workforce. Fewer people and more AI is the secret formula to follow, many will deduce.
The devil’s in the detail with this one. What is Buzzfeed planning?
It will work directly with OpenAI to create new technology
That technology will be used to “create quizzes, help with brainstorming, and assist in personalizing content”
Buzzfeed says it will not focus on “low-quality content for the purposes of cost-saving”
That’s rich coming from Buzzfeed, I know. As far as I can tell, they are one of the prime beneficiaries of clickbait content.
This is also what the Buzzfeed CEO said to his staff. He’d be crazy to say otherwise to that specific audience.
That’s what they’re doing. Why does Wall Street love it so much?
Because generative AI is at the peak of its hype cycle. The stock market is just herd mentality writ very large.
There is an assumption that this will “turbo-charge” Buzzfeed’s output at a lower cost.
Given that I do believe in the potential of this technology, you might be wondering why I’m no sceptical of Buzzfeed here.
But you get what you incentivise, and we choose to incentivise these sugar rushes over the nourishing fuel of long-term strategic thinking. The lesson for other businesses is to hurry into similar, superficial announcements to drive up their stock prices. That’s what CEOs are paid to do.
The damage is all the greater because we are hindering the progress of technology that could, with the right incentives and careful thinking, truly enhance our creative output. I am sceptical that Buzzfeed, of all publishers, will apply the level of rigour required to make this work. If it fails, it will deter others from trying.
The press reaction to the Buzzfeed announcement has been negative. One would expect this: journalists at other publications would hardly celebrate what could be a threat to their profession. But beyond the self-interest, they do have a point. We oversimplify the journalist’s work if we believe AI could replace them so soon.
Every few years, a new technology comes along and lots of people ask me if this technology will steal their jobs. The answer was no then and it’s no now. I ask people if these historical technologies decreased their workload in any way. The answer is typically that they are working more now than they ever were.
It would be a sign of progress if generative AI started doing some of our work for us. The way Buzzfeed is going about it will create more work for the fewer staff that remain.
CNET has started using AI to write articles, which are then edited and approved by people.
In other news, this happened this week:
Some would breezily bat this story aside, saying blithely that the tech will improve over time.
And it will - at a cost and with human effort. Buzzfeed’s stock didn’t jump on the promise of faulty AI that requires significant human correction >50% of the time. The traders bought into the vision of AI creating “endless possibilities” ( that’s Buzzfeed’s phrase).
Now I’m not much of an expert in anything, as you know.
I was permanently deterred from the notion of considering myself an expert by a documentary about Carl Jung I watched many years ago.
In his later years, Jung was at a luncheon and everyone was queuing up to laud his genius. Jung pointed to one of his own fingernails and said slowly, “Everything I know for sure to be true about this world could be written on there.”
If Jung thought that, then the average clown off LinkedIn would need the James Webb telescope to locate their corpus of knowledge. I’d need binoculars, I reckon.
With that in mind, what right do I have to challenge the collective wisdom of Wall Street on my own?
Well, I try to make up for my lack of expertise by speaking to quite a lot of people in different fields.
For instance, I speak to people in AI who have experience and qualifications and such. They are excited about generative AI and cautious about its limitations. There is a sense that this is a delicate moment, when so much promise can lead to disappointment if we don’t put the right structures in place to support the new wave of AI.
I speak to digital transformation specialists who all say that AI (or tech more generally) should be used to create more than just efficiencies. Now is an opportune moment for businesses to re-examine the problems they exist to solve and use AI as a means to deliver value in new ways.
That sounds great, but it requires a leap of faith. It requires investors to go into the unknown, when instead they can make a quick buck from what they know will work.
Meta, Google, and Amazon have all learned lessons on this front. Amazon’s Alexa investments have not paid off as expected, while Google’s moonshot plans have never led to products that can rival Google Ads for revenue. Zuckerberg bet it all on the metaverse and so far, it’s not looking great.
Stock buy-backs look like a better investment than untested AI business models.
Jean-Jacques Rousseau said “take the course opposite to custom and you will almost always fare well.” That wasn’t true then (Rousseau’s own intractable nature caused him much suffering) and it’s not true today.
There’s another group of people I speak to: equity traders. When they explain their work, there is one phrase that appears rather often:
“It’s really complicated. It is. But at the same time, it’s not complicated at all.”
Taking the course of custom is a sure-fire route to success. The trick, then, is to alter this course by rethinking our incentives to encourage investment in the productive use of AI. You get what you incentivise, after all.
See, the EU has fined Google a total of €8.4bn in antitrust cases:
2017: European Commission fines Google $2.73 billion for self-preferencing with its comparison shopping service
2018: EU antitrust regulators fine Google $4.3 billion for requiring smartphone makers to bundle and include the company’s apps with Android
2019: EU fines Google $1.49 billion for making unfair demands of publishers that sought to use its AdSense for Search service
But Google will accept these as merely the cost of doing business.
However, the US is starting to take a closer look now:
In 2020: A coalition of 10 states, led by Texas, file a complaint against Google for maintaining an illegal monopoly over the online ad business. The Justice Department also accuses the company of maintaining an illegal monopoly over search by signing massive deals with partners like Apple and taking other steps to reduce competition.
Google finds itself fighting on a number of fronts. The US government is serious about splitting up its advertising business.
What’s more, the government has its own axe to grind. It cites its advertising spend of $100 million for US Army recruitment as evidence of Google’s illegal practices. The action provides details of that ad spend and the ways Google rigged the ad ecosystem to keep as much of the budget for itself as possible.
In addition, Apple will launch a new search engine to reduce its own reliance on Google this year.
Google contends that Apple’s nascent search business, plus Microsoft’s OpenAI partnership and TikTok’s rapid rise, are all ample proof that it does not have a monopoly in digital advertising.
The problem for Google is that governments are so slow to act. The US is analysing Google as it was a few years ago, not how it is today. Its practices are still as murky as they ever were, but its market position is under threat for the first time in over a decade.
Twitch has shared a road map for features it’s working on for 2023. Among them are changes to the way ads are run, new first-party tools to help with chat engagement, and tools to help streamers manage their brand deals. But the biggest is making pre-roll ads less annoying.
And they’re going to make it easier to clip up streams and share them on TikTok.
This is big news in the search marketing industry. These ranking algorithms are very carefully guarded and we rarely get a glimpse into how they work, never mind a full look at the details.
I’d be wary of transferring these findings from Russia’s dominant search engine to Western alternatives, however. If anything, this might tell us how Google worked before it went “machine learning-first”.
No point resisting the inevitable, eh? Meta is also set to work with OpenAI and Shutterstock.
Similarly, Google has announced that it no longer seeks to ban “AI-created content”.
And to round us off with a big laugh, this bear triggered a wildlife camera and ended up taking over 400 selfies: