Psycho Killer Qu’est-ce que c’est ?
As if to prove the point of how messed up humans can be, some MIT researchers decided to see if they could create “the world’s first psychopathic AI.” To train this machine learning system, the researchers submitted it to a pool of data curated from the darker corners of Reddit. Now, where a ‘normal’ AI sees “a couple of people standing next to each other” in what is clearly a picture of a bat-pigeon, Norman, the psychopathic AI, sees “pregnant woman falls at construction story.”
Nice work, folks! Actually, I think stuff like this is useful because it clearly shows just how much of a reflection of ourselves our synthetic intelligence is. If we’re not careful, we can, indeed, create a monster.
Emotional Machines
Speaking of the psychology of machines, Richard Yonck has a thoughtful piece in Psychology Today on the biological constraints on infusing emotions into machine intelligence.
Perhaps certain types of neural nets, such as generative adversarial networks (GANs), could one day be trained to mimic the triggers and behaviors of an endocrine system? I don’t know this with any certainty, though I suspect in time something like this could be feasible. Nevertheless, these would still be very different from the chemical messengers that humans rely on and so would only be approximations of how our own minds and bodies respond to external and internal conditions.
Losing Control of Our Time and the Rise of Bullshit Jobs
David Graeber writes about how people came to lose control over their time with the rise of time-tracking instruments and modern notions of work.
A bullshit job—where one is treated as if one were usefully employed and forced to play along with the pretense—is inherently demoralizing because it is a game of make-believe not of one’s own making. Of course the soul cries out. It is an assault on the very foundations of self. A human being unable to have a meaningful impact on the world ceases to exist.
This interview with Graeber at Vox is also good.
You Smell Sick
Researchers are developing technology that will be able to detect disease from your breath. The team at Loughborough University in the UK is using deep learning techniques to interpret results from gas-chromatography mass-spectrometers and identify the presence of particular molecules. Their immediate focus is on detecting aldehydes, which can signal the presence of cancers, but there are many other potential applications.