Will Taylor Swift kill off open AI?

Welcome to AI Collision 💥,

An image of a young woman with striking resemblance to a blonde, slender celebrity, in a spirited pose at a Superbowl event. She's dressed in a chic, contemporary outfit, possibly a stylish dress and a light jacket, exuding a glamorous yet casual vibe. The background vividly showcases a Superbowl stadium, brimming with enthusiastic fans, bright lights, and iconic Superbowl banners. The woman is smiling broadly, capturing the joy and excitement of the event, with her hands up as if cheering. The scene should be lively, colorful, and reflective of the Superbowl's energetic ambiance.

In today’s collision between AI and our world:

Zuckerberg is no avatar fool
It takes a Swift move to change AI regulations
Which sauce?

If that’s enough to get the A100s and H100s computing, read on…

AI Collision 💥 Facebook < Meta < MetAI?

On 14 November 2023, I wrote to you about the huge number of Nvidia graphics processing units (GPUs) that Elon Musk needed to buy to increase the computing power of his Grok AI platform.

As I wrote:

On a podcast recently Elon said,

“Grok was trained on 8,000 A100s running at peak efficiency and Grok’s going to get a lot better, by the way, we will be more than doubling our compute every couple months for the next several months.”

The A100s he’s talking about are Nvidia’s A100 GPUs. These A100s vary in price, but there’s a rough estimate that they cost around $10,000 per chip.

So if Grok is trained on 8,000 of them…

You get the idea.

I also toyed with the idea that if Elon was buying up that many AI GPUs, what about everyone else?

Well, we’ve got a bit of an idea who else now.

By the end of this year, Meta’s computing infrastructure will have somewhere in the vicinity of 350,000 Nvidia H100 GPUs.

Last October, Meta founder and CEO Mark Zuckerberg said that, “AI will be our biggest investment area in 2024, both in engineering and computer resources.”

Looks like he wasn’t lying.

How much exactly does all that cost.

Well, I don’t know for sure the price point that Meta is paying. I’d assume it would get a tidy bulk discount.

But you can (try) to find H100 GPUs online if you’re buying just one or two, in the region of $40,000 or so.

In other words, Meta will be spending billions on the levelling up of its computing (read: AI) infrastructure in 2024.

That makes good sense for a company that is very much focused on the development of the Metaverse as a key pillar of the long-term future success of its company.

It also goes to show that Meta is very much clear on its direction. And it’s all about the development of AI and the Metaverse.

It can’t get much clearer than that. Of course, that does mean advertising too – but that’s a given for Meta.

This focus, and the fact that it has spent a big part of the last 18 months trimming the fat, so to speak, has changed the company’s fortunes around.

Not that it ever really struggled, but Facebook was ridiculed, if you recall, for the idea of changing its name and going “all in” on some pie-in-the-sky future Metaverse thing.

The New York Times wrote,

“Shifting a 68,000-person social networking company toward the theoretical metaverse has caused internal disruption and uncertainty.”

But here we are, about two years down the track, and Meta has turned into one of the best-performing mega-caps of the last year.

So, maybe Zuckerberg is onto something?

Maybe this 2024 dedication to amping up its computing and infrastructure with a distinct smell of AI about it, and, of course, the Metaverse is something to keep a closer eye on and, perhaps, take greater consideration for a portfolio.

AI gone wild 🤪

Taylor Swift is arguably the most famous person alive today.

Over 280 million Instagram followers. More than 51 million subscribers to her YouTube channel. Over 70 billion streams of her songs on Spotify.

Yep, she’s got a thing or two going on.

Now add to the mix that she’s dating an American Football (NFL) star for the Kansas City Chiefs – which means she’s very much in the spotlight at this time of the year.

The reason being she’s attending every Chiefs game, and now they’ve won the divisional title and will be in the latest iteration of the Superbowl (the most watched and most popular sports event in the US). The “internet” is already calling this the “Taylor Bowl”.

However…

As great as the NFL’s social media engagement has been since Swift became a regular staple at Chiefs games, there’s a darker element that’s also brought Swift into the spotlight again.

The use of AI to generate malicious images of Taylor Swift in some pretty horrible situations.

This isn’t the first time I’ve spoken about this. This isn’t the first time this has happened. But if anything speaks to the reach and power of Taylor Swift, it’s the fact that the US Congress is debating AI controls only now because of the viral AI images that were created.

Social media sites have begun to block searches for these images and you’ll struggle to see them now (hopefully). But this has accelerated calls for tighter regulation and more stringent and strict laws around the creation of deepfake images, videos and harmful AI-generated content.

Again, none of this is new.

None of it is acceptable either.

The use of AI in these ways, however, must be understood at the root of the problem.

The issue here isn’t with AI. The issue here is with humans.

This is a humanity problem, not a technology problem.

So while the first reaction of many is to look at this as an AI problem, and that it was AI that generated these images. This is not the problem at all.

Any of these images could have been created on Photoshop or any other image creation software. In fact, deepfake origins go back to before the turn of the century.

And the AI tools that are used to create images like these today all require a significant amount of human input to get the AI to perform the task as intended.

This is not a case where rules and regulations need to be harsher and tighter on AI and the technology. The answer to the Taylor Swift images and any other malicious and hurtful content creation is more painful and punitive punishment for those who create it.

Boomers & Busters 💰

AI and AI-related stocks moving and shaking up the markets this week. (All performance data below over the rolling week).

man in black suit jacket and black pants figurine

Boom 📈

Cyngn (NASDAQ:CYN) up 95%
Duos Technology Group (NASDAQ:DUOT) up 38%
Appen Ltd (ASX:APX) up 16%

Bust 📉

iRobot (NASDAQ:IRBT) down 14%
Tesla (NASDAQ:TSLA) down 11%
Guardforce AI (NASDAQ:GFAI) down 7%

From the hive mind 🧠

AI has become such a buzz topic that an hour-long George Carlin “AI special” claiming to be AI just to help it go viral in reality isn’t AI at all.
Trust is everything when it comes to information. In a quest to regain trust in the news, more stations are turning to AI. But would you trust AI news over regular news?
The uncanny valley is something I’ll write about more soon. But until then, the question is: when is real not real, and when is not real, real? Google’s latest release hopes to confuse that question even more.

Artificial Polltelligence 🗳️ the results show

Tangible investing vs. somewhat-intangible investing. This was the idea behind last week’s poll, when I asked if investing in AI hardware made more sense than investing in AI software.

And the results are conclusive…