Monday

So, I already played the Moral Machine last week. I think that programmers should allow the AI to run thousands of simulations of every possible scenario until it learns to find and use the one with the best possible outcomes for humans. This would be the best way to keep bias out of the program. The more our thoughts and beliefs we put into the program, the more likely the program is to be biased towards others.

Joy Buolamwini was fascinating to watch because I had already seen her in Coded Bias on Netflix. However, it is hugely concerning to see how an AI facial recognition program can better recognize a white mask as a person than an actual person of color. The main takeaway from her talks has been that when most of these AI white men program programs, there will be bias in the programs towards people of color, even if the programmer did not intend to do so.

Tuesday

I explored the White Collar Crime Risk Zones project. White-collar crimes are as severe and have life-changing consequences as their street crime counterpart. Although street crime is far more violent than white-collar crime, white-collar crime has a massive economic impact and severe consequences for everyone involved and targets. White-collar crime is also more prevalent. So why do AI predictive crime programs only focus on street crime?

I think the answer is related to two facts. First, white-collar crimes are committed in private rather than street crime, which, as the name suggests, is engaged in public—that reason I can accept. The second fact is that most people who commit white-collar crimes are caucasian—that reason I cannot get it. Not focusing on one branch of crime because of who mainly commits those crimes is horrible, unethical, and just plain racist. Since white-collar crimes are committed privately, wouldn’t a program that predicts where white-collar crime is likely to be committed be invaluable to law enforcement agencies?

Friday

The AI I collaborate with most days is the predictive text function on my iPhone.

AI and creative artists are a fascinating combination. It is interesting to see what a computer generates when you give it a prompt to visualize. However, I do not think the art world is ready for AI art to be considered real art. Instead, the art community is divided on whether art made on a computer should be regarded as real art. Unlike physical art, where mistakes rarely can be fixed, the artist can permanently correct errors in digital art. It also requires less time and tools than physical art, leading some artists to say that digital art does not require any skill or talent. While more and more people are beginning to appreciate digital art made by humans, it will be long before AI-generated art gets that same appreciation.

If AI programs reflect the human/humans that programmed them, then a computer can be creative. However, it may never have the capability to be as creative as humans since it is abiding by lines of code, but it can certainly come close. Unfortunately, not all humans are creative, and some are more creative than others. Perhaps if people with innovative and artistic talents in the liberal arts have some say in the programming of AI, then these programs maybe just a step below them in terms of creativity.