Let’s come to an understanding of how we got to this moment in the evolution of artificial intelligence. Similar to the way scientists controversially study and tinker with viruses — under the assumption that they’ll naturally mutate anyway, so we should try and prepare for them ahead of time — a few unaccountable masters of the universe not that long ago decided that the world was probably headed in this AI-focused direction anyway, so why not hurry the process along?

That, basically, is the thought process that let the genie out of the bottle and gave us companies like OpenAI, which is now worth $86 billion and is led by a boy king with vocal fry who believes there is such a thing as a “median human” and who has yet to be candid with the world about why OpenAI fired then rehired him. For better or worse, his company is now at the vanguard of a race to develop products that are as good or superior to humans at a growing number of tasks.

Thus, the race began. Organizations from Silicon Valley to Wall Street, scared to be left behind in the scramble that OpenAI started, developed a case of AI Tourette’s and started rolling out their own AI strategies. This is why search giants like Google and Microsoft are racing to give control of the internet to AI chatbots that are essentially automated plagiarists at scale, eating and then regurgitating other people’s content — while telling those of us who produce that content to do so “with people in mind,” and not the machine that’s going to either steal that content or make it so that no one can find it.

To that last point, HouseFresh (a reviews site focused on air quality products) recently went viral with an article about how Google’s actions are killing small, independent publishers. As a measure of the extent to which Google is now garbage, that article — titled “How Google is killing independent sites like ours” — had previously been outranked by sites like Reddit and by other publishers writing about the HouseFresh article. Now, though, after a massive Google update that’s rolling out now, that HouseFresh article doesn’t even rank on Page 1 of Google Search, which is populated by other sites writing or talking about that article.

If you think the search giant that built the system behind that ranking methodology is going to somehow add an AI chatbot to its search experience that makes that whole situation better, I have a bridge to sell you.

Even the venerable BBC is not immune to all this, with the Beeb now saying it will use AI to craft marketing and promotional communications for its long-running sci-fi show Doctor Who (to the consternation of fans, who rightfully recoil at such a practice for a show wherein AI is always, and rightfully, portrayed as an evil to defeat).

AI is going to improve so many things, they tell us, not unlike the way technologists once propagated the absurd notion that giving everyone in the world a microphone, via Twitter, was a de facto good — notwithstanding the fact that there are a lot of crazy, evil, and mendacious people in the world. This time, though, what’s happening is not something that can be scoffed at, the way contrarians used to dismiss Twitter as little more than a vehicle for telling the world you ate a sandwich.

In fact, there’s a profoundly disgusting anti-human sentiment nestled at the heart of the AI fetishization happening all around us. It’s a push that, quite literally, amounts to a natural extension of late-stage capitalism and is largely (but not completely) about replacing people.

Think about all those realistic-looking videos that have gone viral from OpenAI’s Sora. They certainly look impressive and incredibly realistic … but it’s not like we haven’t encountered computer-generated imagery before. The difference between a Pixar movie and these OpenAI videos, obviously, is not in their quality — rather, if you find yourself blown away by the latter, all you’re really letting yourself be amazed by is the speed with which those videos were created, and how little effort it required.

Which is another way of saying it impressed you to not need humans anymore.

In many respects, it sort of feels like Silicon Valley broke the world and now is just saying to hell with it, machines can build a better one anyway. Here’s another example of what I mean about the AI push feeling deeply anti-human: Altman is reported to have said that the AI industry needs to raise a few trillion dollars to make the necessary investment into chips that will allow the industry to go to the next level. And if the past 24 months have taught us anything, it’s that what Altman wants, he gets.

One can only imagine the far-reaching and compounding good that would result from using those trillions of dollars, instead, to fix the country’s crumbling infrastructure, schools, and health care system. The money is obviously there. But the privileged few who already live lives walled-off from the rest of us have decided that money would be better spent on things like AI image generators that get testy when you ask them to draw a picture of a white person and programs that can arrange a spreadsheet better and faster than you can. I can’t say I blame the crowd at SXSW on Tuesday who loudly booed when OpenAI VP of consumer product (and ChatGPT head) Peter Deng made the following statement:

“I actually think that AI fundamentally makes us more human.”

[ For more curated tech news, check out the main news page here]

The post AI companies like OpenAI are basically performing gain-of-function research on humanity first appeared on

New reasons to get excited everyday.

Get the latest tech news delivered right in your mailbox

You may also like

Notify of
Inline Feedbacks
View all comments

More in computing