(www.theguardian.com)
What a time to be alive
Last time that AI was big before DL it was the "big data" fad and everything had to be big data. Marketing has never not been about how to disguise "what we already do" as the newest buzzword that customers (or investors) want to hear.
The same goes, of course, for all the non-AI fads like "the cloud" or "NoSQL".
I remember in the years before the pandemic that I would joke that all you had to do was "sprinkle in some blockchain" to your VC pitch and your valuation would automatically go up by tens of millions. It seemed dumb to me then and it seems dumb to me now.
The people who are being marketed to with the AI term don't have any idea what that mean and AI, as a marketing term (the only way it's ever been, so far, commercially used) means a lot more than transformers. My dishwasher has "AI" because it has sensors that can detect where the most dishes are.
The marketing term really just means that the product changes it's behavior without user input. A simple "if...then" is AI.
AI has been used as a marketing term for at least a decade now but LLMs are poisoning the brand because they're, largely, implemented in almost exclusively user hostile ways.
Slightly ironically, now in the mirrorless era, and AF algorithms actually based on DL subject recognition and complex predictive algorithms, Canon has retired the "AI" label.
9i - "internet"
10g - "grid"
11g - "also grid"
12c - "cloud"
26ai - "ai"
various other examples. One really annoying thing is this has also happened in open source projects too - generic things that, sure, help out with AI tasks are now "AI" things.
They’re incentived to do so because apparently investors don’t understand the difference.
If the market can remain irrational longer than a fundamentals-driven investor can remain solvent, is it irrational to bet on the market remaining irrational?
Remember the Internet was first hot and everything was iThis or Active That. iPhone still has i.
Remember… well not, me, I wasn’t alive… when radiation was cool and Radioactive was in.
Everyone always wants to be cool.
The reason the term "machine learning" was even invented was because it was one of the AI winters and an euphemism was needed because "AI" was more of a swearword than a buzzword.
which used to be called Statistics
which used to be called “math” or maybe “applied science.”
Obviously the underlying tech and research changed along the way… but not as much as it would seem. We’re still doing matrix operations and gradient descent and softmax, all of which has been around for a while.
Last time around was when "fuzzy logic" came out, I think?
If negative perception towards AI grows, either because of negative experiences after having it forced on them, or as people's utility bills skyrocket, or as the environmental impact becomes more apparent, they might find that what appeals to shareholders doesn't impress the people who usually pay for their products.
I don't think people understand just how viscerally negative the perception of AI is for the youth.
I'm doing Quantum Crypto AI next.
> Any sufficiently advanced technology is indistinguishable from magic.
In the sense that it doesn’t matter if it’s AI or Algorithms. All that matters is people think it could be AI. If yes, then it is. Doesn’t matter what’s actually going on behind the scenes.
(Not that I love this reality, I don’t advocate for it, but this is how things are)
One of the sample prompts is "How is my traffic compared to last month?" So if you type all of this text, click send, wait for Google servers to burn a liter of water to calculate a probable answer, what it gives you isn't even the answer, but an option that you can "apply." If you click on "apply," it refresh the page with a filter using the functionality that already existed in the search console. In other words, this entire LLM can't do more than you can already do by clicking on the extremely simplified buttons of the existing UI. How do you do the same thing via the UI? Click "More -> Compare -> Apply". A whole LLM to replace 3 clicks with 2 clicks + typing the prompt.
By the way, just think: if we gave people an LLM in this analytics thing, what is the number 1 question people would ask? The answer is obviously "how do I increase my clicks?" or "how do I become number 1 on Google?" You don't even need to be a product person to figure that out. That's obvious. Just completely obvious. And of course, the Google's chatbot can't answer that. Because they probably realized, instantly, that is going to be a lawsuit if they said "do X to get more clicks" and you did X and you didn't get more clicks.
Layoffs aren't indicators of success or failure, just some theoretical tea leaf reading style signal for future profits of a company. So if a company is already growing unbelievably fast layoffs are a bad sign, if a company is slowing in growth apparently having employees on business units not working out is equally a bad sign.
And for many companies these layoffs are the modern version of Roman public executions with the audience(investors) cheering it on.
They didn't trade company fundamentals, they traded the market sentiment.
In actuality AI is the consultant.
I like AI, but seriously, who actually invests on this basis? Where is the critical thinking? I don't feel sympathy for any investor that gets rug pulled on this stuff.
No one cares about the product any more. And that will be the end of all of this.
The more trendy boxes you tick, the broader the universe of people whose box you tick and who can thus invest.
This is one of the reasons stock market is so disconnected from reality.
Perhaps the investment is more on the “greater fool” theory. “I think this is complete nonsense, but there’s probably someone not as savvy who will buy into this garbage idea upon which I can profit.”
They don't and the people who are falling for this rhetoric are naive. Most investors _should_ invest more in AI companies. And most companies _should_ invest in AI. It is the rational move and it is exactly what we are seeing here. I don't know what the hysteria is about.
At the height of COVID, food photography was very important. Because of distancing requirements and his kids health, he didn’t really have access to hiring photographers and so he invested in a good camera and a tripod, and started to learn to be restaurant’s photographer. Six years later and he’s still the photographer but he’s back to using an iPhone and he’s forgotten a lot about composition because obviously not AI generated has become a differentiator.
A while back we ran out of .com domains and that burst the bubble. Or something like this.
”I've known men who inspire fear. Do you know what they have in common? They never say how frightening they are.”
And here we are.
”I’ve known companies that work on AI. Do you know what they have in common?...”
A genuine gangster doesn't feel a need to flex their power, because such gauche displays only highlight one's own insecurities and weaknesses
Especially those who have not implemented software in businesses trying to suddenly boil the ocean with AI.
AI remains a great step forward to help businesses benefit from technology, with more than one competency around the table.
The article gives three examples
- Allbirds, a shoe company
- A genetics company marketing that it is using AI
- a property tech company using AI to create 3rd landscapes
The Allbirds one is just financial re-engineering. The others are reasonable?
In the same paragraph as the genetics company, they also mention an "AI-powered basketball hoop" and "AI-powered lasers that – somehow – protect women from predators on crowded underground platforms." Very reasonable stuff that you forgot to mention.
This is important not just for cynical reasons, but to calibrate exactly what it means when we look around and see that "everyone" is using AI these days.
To think otherwise is naive.
All those threats of "maximize token spend or else something unspecified horrible happens to you in the future" are super weird.