Ai Agents will be coming for your Internet. And everyone else's. Will you be ready? Can you be?

We don't have to wait for SkyNet. We don't even have to wait for AGI. The generative AI Large Language Models that we have right now are plenty powerful enough to wreck the Internet when combined with existing social media, ubiquitous video, and a twist.

There's about to be a war. Not a "Robots Kill All Humans" war, but an "iPhone vs Android" war: and it's for who gets to have the winning AI Agent. And what those Agents will be able to do... They might not be as scary as the Agents from The Matrix, but they'll be bad enough, and they're a billion billion times more likely.

And if the Agent's capabilities weren't sufficiently frightening already - in a race to be first to market, the winner is *always* the Bugs. Careless people write exploitable code, rushing makes people careless, and incentives make people rush. If we're not careful, this is going to be a nightmare...

00:00 Intro
01:37 Channel Intro
01:58 List of components
03:52 Phase 1: Al Agents
05:21 Phase 2: Add Video
07:15 Phase 2b: JOI
08:04 Phase 2c: Checkpoint
10:17 Phase 3: Don't Google This
11:33 Phase 3b: Enter Biometrics
12:16 Now with even less reality
13:14 Viral Conspiracy Hallucinations
13:54 The two corners
15:22 In the race to be first, the Bugs Always Win!!
15:53 A bright side? Maybe not?

UPDATE: I did't see this until after I posted this video, but this dropped in the last couple of days (assuming it isn't faked):

Things are moving fast...

# Links from the video:
## Warning about AI Agents

# Rise of Video

# ChatGPT leaks data if you ask it to repeat stupid things

# Cool game with blinking as a control scheme

# AI-generated video

# CAREFUL WITH THESE: You can't unsee or unlearn what you might see or learn here.
# GREAT reporting - disturbing topics:

[InternetOfBugs]

Like with lots of dual-use tech, there are upsides. The accessibility uses are real. But it's a whole other way of profiling what users are looking at and optimizing for keeping you on their platforms...

[hblaub]

Imagine an AI virtual agent gaslighting you into buying more stuff or after a hacker implanted a trojan into it

[aaronbono]

A friend of mine asked me this week how to tell the difference between something that's real and AI generated. I told her it's going to be more important than ever to meet people face to face.

[TheHumaun]

Dear Carl!

You always impress us, your audience, not really by your talents or skills, but by something deeper, far more important. As we see there are many things in the internet and the world that is increasingly selfish, profit-seeking and short-term oriented. You are driven by a higher cause. You, as I perceive, recognize the danger of being reckless in pursuit of petty things. You can't sit there silent and you started this channel to open people's eyes.

I respect you profoundly. You are true to your movement: you don't seek hype, profit - you think bigger...
Together, I believe, we can change the world and move Humanity forward!

[eemelikyrolainen]

I feel like one of the scary parts of these AI agants will be other companies buying "influence" on the agent and thus making the AI agent steer people towards their products.

[cherubinth]

TikTok but all videos are AI generated, the AI learns from the users watch times and swipes to make it as addictive as possible.

[dnmartinz]

I respect your content really much, sir ! i came with the Devin exposed but turns out was an amazing discovery.

[nask]

I am glad to have found this channnel (just recently). Great content man, thanks.

[gank]

This happened in the 2010s with the death of the internet theory, it’s only going to get worse year by year.

[docopoper]

At least Linux and GDPR exist. I've already told Google to delete almost all of my data and am running Debian. Microsoft is heavily pushing AI integration into Windows and I have no intention of ending up in that ecosystem. Also, I'm actually an AI researcher working at a university.

[DigitalyDave]

Lol! That intro/outro is pretty awesome! Thanks for the content my dude!

[blwatson]

I'm in...my man made a Voltron reference

[happyaccidents]

Finally a video on the dangers of whats coming with actual factual backing. Not fear mongering these concerns seem very plausible.

[BrianMartensMusic]

When Black Mirror doesn't go far enough...

[PowerGurhl]

Love your channel! I hope you keep posting. I know more people will find you and find your videos interesting. They are gems!

[MrVoidfull]

Well the Humane AI pin is a perfect example of what you're saying. Time to market completely took over code quality (and probably security). We're just lucky right now that the quality is so bad, and the price so high that the product is unusable. That's not going to be the case 10 years from now. Just like any engineering product, people are going to be building on top of this until there's going to be a usable consumer product. Code quality and security will overall be the same, with enough time and work thrown at it to make it usable for consumers.

As for all else presented here, that's why I don't want to use any smart watch, smart tv or smart assistants. Trying to keep my devices as dumb as possible. That's going to be harder (or impossible) with time, especially since I'm a SE, working with copilot is genuinely great, and using an iPhone vs an adroid device with Graphene OS is also pretty great. But with what's to come, it may be wise for me to step down a bit an get back a bit more control over the things I'm making use of and how I'm using them.

[WallaceRoseVincent]

Thanks Carl. Im watching all your videos for a good update on the industry. Appreciate your point of view.

[JoJoDramo-ihqk]

Finally, the capacity to generate infinite amount of custom made semantic noise to overfit any algorithm to maximum dopamine output.
Not very sure if there is much to extract anymore, but ai finds a way. bless.

[m]

What frightens me with AI is how it can exploit the gullibility of the average user... given its conversational format and the fact it provides an answer, even if it's wrong or fictional... people seem to take bad answers from AI a lot more seriously than they should... it'll be just another marketing tool soon enough.

[JerkyTreats]

Funny thought experiment with personal agent: MFA auth can just be requested to your agent, because it knows where you are, what you're doing, at all times. User interaction not required.

The thought experiment is just how terrifying that is, especially if a malicious actor gets access to your agent.