robot


2ND ENTRY.

  • Can we program morality into robots?

Tuesday. March 28th, 2023.

There are many philosophical issues posed by emerging technology and AI. Machines will eventually have to make more ethical choices as they become more and more integrated into human's lives. You will make a lot of ethical decisions throughout the course of a day, many of which will be based on your personal values. Do you give up a seat on a bus? Do you drink the last of the coffee in the office kitchen? Do you cut ahead of that person on the highway? Are you caring, faithful, and honest?
Thus, teaching morality to machines is fairly challenging because humans themselves cannot objectively convey morality in measurable metrics that make it easy for a computer to process. In reality, it is debatable whether humans even possess a solid morality that can be embraced by all. Humans frequently rely on intuitive instinct in moral judgments rather than complex cost-benefit analyses. On the other hand, machines require specific, objective measures that can be precisely measured and improved. For instance, an AI player can succeed in games with defined rules and limitations by repeatedly playing them and learning how to maximize the score. However, problems with efficiency are a lot trickier in the real world. To give an example, how can you program a computer to deal with gender and racial biases in its training data or to algorithmically maximize fairness? Without a clear definition of what fairness is, engineers creating an AI system cannot teach a computer what is fair.
Due to aforementioned reasons, some authors and programmers are concerned that the naïve application of algorithms to prevalent problems could amplify systemic discrimination and replicate biases in the data that these algorithms are built on. All in all, we must ultimately define morality and measure it in clear terms given the status of AI research and its uses in society. Although challenging, this job is not impossible to overcome.


—YOSANO AKIKO.

Edit
Pub: 28 Mar 2023 12:24 UTC
Edit: 28 Mar 2023 12:26 UTC
Views: 64