This week, we’re thinking a lot about intelligence. Is a brain still a brain if it’s uploaded to a computer? When a robot paints a portrait, does it count as art? How do dogs know the difference between a cry for help and a hum?
If your computer only had a brain
Scientists are working on a way to create functional copies of the human brain on computers. The technology is still in the developmental stage, but futurists like Ray Kurzweil believe that people will be using this technology to become “digitally immortal” by 2045.
But how does the idea of digitizing the human brain strike you? A research group at the University of Helsinki is trying to gauge how ordinary people feel about the practice. The group, Moralities of Intelligent Machines, led by Michael Laakasuo, collected data from people in the US.
They found that more men than women expressed approval of the technology. But, when they looked at the numbers for science fiction enthusiasts, the gender differences evened out. They also found that people affiliated with traditional religions reacted negatively to the notion of mind uploads.
This research is part of a greater effort to understand the moral psychology of robotics and artificial intelligence (AI). We may one day be faced with difficult questions such as, “can the digital brain feel pain?” and “does powering-down the digital brain amount to homicide?”
Speaking of machines taking on human characteristics, what about paintings made by robot artists? Works submitted to the 2018 RobotArt competitions were impressive.
CloudPainter won first place with Cezanne-inspired portraits and landscapes. The second-place team presented Impressionistic paintings that demonstrated expert-level brush strokes. The third-place entry used haptic recording and playback to create reproductions of Van Gogh landscapes.
The robot paintings are good, but human artists shouldn’t worry. RobotArt founder Andrew Conru says that art created by people will always be valued because it offers a shared human experience. But what if that robot artist has a copy of a human mind?
It’s not such a small world
We use social media to connect with people all over the world. If you live in the US, you might have Facebook “friends” in far-off places like Zimbabwe and Japan, but your global interactions are probably infrequent.
Professor Ming-Hsiang Tsou and alumna Su Yeon Han of San Diego State University led a study in collaboration with Keith C. Clarke, a geography professor at the University of California, Santa Barbara. They focused on users’ online and real space interactions in four US cities.
The researchers found that the chances of being real-life pals with your online friends decrease as geographic distance increases. Interestingly, it was also discovered that your experience with online shopping improves if the purchases can be shipped from within your geographic region.
Dogs to the rescue
We’ve covered machines, and people; now let’s look into research about the family dog. A study conducted by Emily M. Sanford, formerly of Macalester College and now at Johns Hopkins University, showed that dogs can sense when their owner is in distress, and they’ll step in and try to help.
Sanford and her team had the owners of 34 dogs sit behind a closed transparent door. They were instructed to either cry out in distress or hum. Both sounds caused the dogs to open the door to get to their owner.
However, the researchers observed that the dogs who heard a cry of distress responded more quickly than the pooches who heard the hum. The distress-responding dogs also remained calm during the test. The researchers believe this shows that a dog's ability to suppress their own stress allowed for a quicker response.
When AI is for the birds
And now from canine companions to feathered friends.
Bird populations are dwindling because of climate change, logging, and agriculture. But scientists can monitor species by recording their unique calls. Humans are best at distinguishing the song of one bird species from another, but now, a crowdsourced project is helping artificial intelligence researchers create algorithms to recognize bird calls.
The Bird Audio Detection challenge was devised by researchers in the United Kingdom, France, and Crete. The group released hours of audio from monitoring stations around Chernobyl, Ukraine, as well as crowd-sourced recordings. Then, humans identified and labeled 10-second clips. Some clips contained birdcalls, some did not. Teams then trained their AIs using the labeled clips.
When the monthlong contest ended, the winning algorithm scored 89 out of 100 on a statistical measure of performance. The high score on a previous trial was only 79. The new algorithm excelled because it avoided labeling sounds that did not come from birds, like insects and rain.
Humans still do better at correctly identifying birdsong, but machines don’t need sleep and they’re not bothered by bad weather. Now that’s something to tweet about!