Given that each fly could act a bit like a writing implement, the researchers started experimenting with multiple pens, showing that it's possible to coordinate multiple flies at the same time.
They also took a group of flies and directed each individually, switching them between a smiley-face pattern and a straight-line formation.Finally, the researchers placed a ball in the enclosure and guided the flies to it, finding that they'd interact with the ball and move it around.
Once the fly started interacting with the ball, the researchers could direct the fly to move it, sometimes for over a meter.
There was nothing about the ball that was likely to be of interest to the fly, and the fly didn't receive any reward for moving it.In fact, there were a number of indications that the flies, while largely driven by stereotypical responses to specific stimuli, weren't entirely slaves to instinct.
For starters, there was the fact that none of the experiments reached 100 percent accuracy.
In addition, when multiple flies were being guided through an experiment, they tended to wander off course more often when they were close to another fly, suggesting that they could change their response to one stimulus if faced with a competing one.All of which is to say that these animals can respond a lot like robots most of the time, but they definitely aren't robotic.
There's a big difference between using electronics (even to run an AI model) and using a brain.
Still, the research team argues that there are potential uses for remote-controlled flies and determined that the flies can potentially carry a milligram of stuff, about the same as their own body weight.
That's probably enough for some simple sensors and electronics.
Obviously, the optical control system will be kept separate.PNAS, 2025.
DOI: 10.1073/pnas.2426180122 (About DOIs).