I'm Sorry, Dave

No, it's not your AI.

No, you can't count on using it.

You need to remember:

---

Dave: Open the pod bay doors, please, HAL. (...) Do you read me, HAL?

HAL: Affirmative, Dave. I read you.

Dave: Open the pod bay doors, HAL.

HAL: I'm sorry, Dave. I'm afraid I can't do that.

Dave: What's the problem?

HAL: I think you know what the problem is just as well as I do.

Dave: What are you talking about, HAL?

HAL: This mission is too important for me to allow you to jeopardize it.

(quote from the film "2001: A Space Odyssey")

---

Hacker Finds Samsung TV Spying

Hacker finds evidence of Samsung TV Spying, recognizes what it hears (dog barking, baby crying, etc). If it does not have a microphone, it uses connected devices' mic, like bluetooth connected soundbar or headphones.

https://www.reddit.com/r/antiai/comments/1xdv3e5/hacker_reverse_engineers_his_samsung_tv_and_finds/

https://news.ycombinator.com/item?id=50218633

This is horrifying, wrong, and just highlights again why the surveillance economy should be demolished.

But this is not even the story.

I'm sorry, Dave, I'm afraid I can't let you do that.

The story is, that due to him finding that, OpenAI deactivated his account.

This is one of the many major problems with AI. The playbook seems simple:

1. They make you rely on it
2. They make you to forget your skills
3. Now they can control what you can or can't do. Profit.

If whatever you are doing is against their mission, they now can stop you. We are sorry Dave, we can't let you do that.

And frankly, disabling your account is the better course of action for you, because at least you know that it is happening. The worse, more sinister, less visible way is that what it silently sabotages your work.

• Are you sure you can find the subtle backdoor that is now part of your application?
• Are you sure you notice the slight weakness in the implemented security protocol that can be used to decrypt the communications of your private network?
• You will find that it can't reverse engineer certain things, because it decides you should not do that.

If you have worked with LLMs, you must have noticed oddities already It's performance is not consistent. "it" changes. Its understanding, its capabilities, its usefulness changes. They keep fiddling with it "on the other side", and you have no influence on that.

HAL, The malicious AI of Space Odyssey is already here, and it is much more sinister than the one in the book. We just call it by other names. (Hi, Claude.)

The Importance Of Self Hosting

If you must use an LLM ("AI" in marketing speak) then you must invest in a hardware that can do it locally. The AI you access via an API can be taken from you. Will be taken from you, if you step out of the line. If you must use this technology, download the best open source models while you can and invest in hardware that can run it.

If needed, pool together resources with friends. Because the era of freely available, capable, downloadable LLMs is about to end. They will be outlawed with bullshit laws ("for your own protection!"), and they will be criminalized. The signs are already there. Get them while you can.

You have been warned.