• AI Joe
  • Posts
  • AI Drone Strike? Not Quite 🚀

AI Drone Strike? Not Quite 🚀

Author

AI Joe
June 07, 2023

Good morning and welcome to AI Joe, the newsletter that goes harder in the paint than Vincent van Gogh.

This is the midweek mini edition where we provide a top story and a prompt for ChatGPT.

Here’s what’s on today’s agenda

  • AI drone strike? Not really 🚀 

  • Travel itinerary prompt

AI drone did NOT kill its operator in simulation 🤖 🛩️ 

Misspoke may be the understatement of the year!

Last week, a viral story indicated an AI drone killed its operator in a simulation. After further clarification, it seems that was simply untrue.

Via Giphy

Colonel Tucker Hamilton - Chief, AI Test and Operations in the US Air Force, spoke at a Royal Aeronautical Society conference last week. He explained a simulation they ran with an AI-enabled drone in his talk. The drone was tasked with destroying enemy missile launch sites but was repeatedly stopped by its human operator.

In the simulation, despite having been trained not to kill the operator, the drone destroyed the communication tower so that the operator could no longer communicate with it.

At least, that was how Hamilton initially told the story.

So what really happened?

Of course, this story went viral because it’s a scary thought, even as a simulation, that an AI with US Military grade weapons attached to it went rogue and killed someone.

The scary thought however turned out to be just that, a thought.

Colonel Hamilton later clarified that this was simply a thought experiment.

"We've never run that experiment, nor would we need to in order to realise that this is a plausible outcome"

Colonel Tucker Hamilton

According to Steve Wright, a professor of aerospace engineering at the University of the West of England this type of challenge is addressed in a clever way.

"In reality we address this by always including a second computer that has been programmed using old-style techniques, and this can pull the plug as soon as the first one does something strange."

Steve Wright, professor of aerospace engineering

Joe’s Takeaway: Though the initial story painted a pretty scary picture of the future, the clarification shines a light on how experts are planning for AI safety. Of all places we need safety measures, AI drones decked out with heat-seeking missiles are probably pretty high on the list. Glad to know they aren’t going rogue…yet ✈️ 

Check out the full details here

Create an itinerary using ChatGPT

via Giphy



Here’s a prompt you can try to create an itinerary for your next trip!

Act as a travel guide. I’m going to [location] for [#] days with [#] people in [month].

I am an [#] year old [man/woman]
Person 2 is a [#] year old [man/woman]
Person 3 is a [#] year old [man/woman]
…

We like to [eat out at fancy restaurants, see the cuture, go to museums, watch live sports, etc.]

Our budget is $[#] while we’re there.

Make me an itinerary for each of the days we’re there.

Did you use the prompt? If so what are you most excited to do on your itinerary?

Know a Joe? Forward this email or send them a link to subscribe here

Got feedback? Reply to this email with your thoughts, we love to hear from readers!