Could military artificial intelligence GO ROGUE?!

A few weeks ago, a story about a drone controlled by artificial intelligence going rogue during a simulation and killing its human operator spread across the internet like wildfire. But while that initial story saw a great deal of coverage, the correction that followed has gone largely ignored.

And for good reason. For those who have long worried that the military’s use of artificial intelligence (AI) would inevitably lead to a war between man and machine as we’ve seen in movies like the Terminator franchise, this story seemed like nothing short of vindication. News outlets and content creators all over the world quickly saw this story as a perfect opportunity to leverage that pervasive fear to drive clicks, views, comments, and shares.

So, what’s the real story behind this tale of artificial intelligence gone bad?

📱 Follow Sandboxx News on social

📱 Follow Alex Hollings on social

Further Reading:

Citations:

[glen]

In other news, the Air Force renamed their AI program Skynet.

[speedisgood]

Came to this channel for videos about the cool NGAD planes and then stayed for smart, reasoned videos about Russian "hypersonic" missiles and now AI. Keep on giving this great content, Alex!

[theodurbin]

This was thoroughly and responsibly done. I would absolutely be interested in a part 2.

[Rapiddrive]

Well done Alex. You’ve taken a complex problem people have loaded emotionally with fear and anger and opened it up logically so we can again see it’s not the sword, it’s the hand holding it we need to fear. Thank you for this.

[sneakerset]

Best analysis I've seen on AI misperception(s) created by media sensationalism for views/clicks/ads. Thank you, Alex !

[Angarsk]

Great analysis. A good friend once said: AI will never self-code, it will always do what a human programmer codes for it. And that, I believe is the basis for what today seems an impossible self-awareness. We don't even know how humans are self-aware, let along program a computer to do it.

[iamcoolerthanconnor]

Hell yeah, this is my favorite part of Friday! Thanks, Alex, I appreciate your insights!

[msamov]

Yes, a Part II Please. IT is always a pleasure to watch your journalism and learn something. The application of the "Eliza Effect" into the AI calculus, as you reported, was a real eye opener. We humans, the great " self-deluders." Thanks again Alex!

[glenn_r_frank_author]

PART 2! PART 2! I agree that today we are not at the place of actually seeing real AI as defined in sci-fi... actual self-aware sapience - or at least sentience ... but the potential that some day soon someone will create that sci-fi form of AI is still there isn't it? Maybe not today, maybe not very soon, but potentially some day.

[googacct]

From the descriptions I had heard about the "rogue AI", it was nothing more than training a reinforcement learning agent. The agent has lots of different options for performing a task and tries different things until it gets the desired result. If you do not tell the agent that doing certain things will result in losing points, then it will try those options. The AI version of better to ask forgiveness than ask permission.

[thefoolishhiker]

That Eliza program ended up morphing into a script that was packaged as Dr. Sbaitso with creative labs sound blaster sound cards in the early 90’s. I had endless fun mucking around with that little program making the “AI” therapist provide absolutely hilariously bad and inappropriate recommendations.

[chrisbrowning]

Excellent video, as always. Definitely want a part 2. This was incredibly interesting and insightful.

[charlessage]

Wow, thank you for this, great presentation. I'm doing research on this topic for a novel and this is REALLY good for the direction the next in this series will go. I've seen a couple of your videos now and I am now a subscriber and automatic like. Keep up the great work.

[raymondtorres-gyuj]

Alex your are a genies brother, thank you for sharing all these super great videos...Keep them coming!! 👍

[weztuppeny]

Hey Alex, A long time fan of your video's from the UK, as the way you go into detail about everything is worthy of a paper as uni people would write it me being one the last few years again. Yes would love to see a part 2 as this is something that is so close to what is probably happening behind closed doors anyway by amazing to see all your points of view and expertise in military aircraft AI etc. Keep up the good work bro!!

[JasonCummer]

Love this channel. Love hearing theory of mind on a video from time to time. Pretty rare knowledge

[WritingRyanEdel]

I would definitely welcome a Part 2 — your analysis gives so much to think about. I especially appreciate how you've peeled away the emotional biases to show that we humans are still our own worst enemies, and AI will simply become another tool to be used for good or ill, albeit one that's far more complex.

[Rattiar]

Nice breakdown. Would gladly watch a part 2!

[agonnoga]

We definitely need a part 2.
Thanks a lot. 👍

[kylecollins]

Yes please add a part 2. I have just subscribed and am now heavily interested in this channel. Any subject on the B21 is a great video for me 👍