Wednesday, May 24, 2023

Cyberethics

Norbert Wiener

In 1948, Norbert Wiener created the term Cybernetics, applicable to a new technology, which he defined as follows:

The study of control and communication in the animal and the machine

Cybernetics has a lot to do with Robotics and with the use of computers and microprocessors to control and communicate; in other words, to do almost everything we use them for.

But what is being talked about right now, rather than Cybernetics, is Cyberethics: ethical issues related to the use of computers, social networks, and most tools that modern technology puts within our reach.

As I have said in previous posts, tools are neither good nor bad: what is good or bad is the use we make of them. From this statement two immediate corollaries follow: the first is worrying; the second, if carried out, could help alleviate that concern. They are these:

  1. Every tool will be misused
  2. It is our duty to try to prevent the misuse of tools

It is obvious that Artificial Intelligence, even in the weak form we have it now, is a very powerful tool. Therefore, according to the previous corollary, its misuse could cause great problems and misfortunes. Let us consider a few:

         Dissemination of incomplete, biased, or simply false information. We have seen that tools like ChatGPT and its successors and competitors do just this. Having been trained with large amounts of information taken from the Internet, as this information is often incomplete, tendentious or false, these characteristics are automatically transferred to the answers they give to the questions addressed to them. In previous posts I pointed out that these tools are useful mainly for those who know the answer to the questions they pose, i.e., for those who don’t need them. Those who need them should not trust the answers they receive, for the tools are programmed to answer something, even if they can’t find the answer.

         Substitution of human workers by machines. This has been going on since the start of the industrial revolution, more than two centuries ago, but some think it could now happen on a massive scale, putting millions of people out of work in a very short time. In an article published in 2013, two Oxford professors, Carl Benedikt Frey and Michael Osborne, predicted that 47% of jobs in the United States are in risk of being replaced by robots by 2030. Although apparently, a few years later, the authors qualified their prediction, this fear has been growing lately, as a result of the new AI tools, some of which are now playing the role of advisor or member of the board of directors of a company.

         Some applications of artificial intelligence increase the risks that threaten us in ordinary life. Research on automated car driving is advanced, but practical progress is slow, not because of technology, but because there is a lack of legislation on the problems caused when accidents occur, which they certainly will. Autonomous weapons, which are being used increasingly, raise the risks of those armed conflicts that, unfortunately, have not disappeared. And for decades now, portfolio management algorithms have been causing undesirable effects on stock prices.

On the other hand, fears that AI applications will become conscious and take over the world, are unfounded. The first may not be possible, and if it were, it would be in the very long term. The latter could happen, but only if we are foolish enough to replace human beings in key positions by computer tools.

But, as the Future of Life Institute points out, AI doesn't need consciousness to pursue its goals, any more than heat-seeking missiles do. This concern has prompted this institute to sponsor a gathering of signatures to request that research on applications such as ChapGPT, GPT4 and their followers and competitors be suspended for six months, while the unfavorable ethical consequences that may be caused by the improper use of these tools are considered. This initiative has collected almost 30,000 signatures. But I doubt that the six-month pause will be enough.

In parallel with this initiative, the CAIDP (Center for AI and Digital Policy) has denounced OpenAI (the company that created ChatGPT and GPT4) before the United States Federal Trade Commission (FTC), accusing GPT4 of being a deceptive product, a risk for public safety, and requesting the suspension of future versions.

These steps may give us more time to consider some of the problems. There is the precedent of research in genetic manipulation, which can also cause considerable risks, some of which still threaten us. Faced with such important issues, we shouldn’t be too daring. It’s better to be cautious.

In his famous book Cybernetics (1948, 1961), which gave its name to the discipline, Norbert Wiener also recommends caution. And he does so by mentioning a horror story written at the beginning of the 20th century by the Englishman W.W. Jacobs: The monkey's paw. After summarizing the story, Wiener ends with these words:

In these stories the point is that the agencies of magic are literal-minded... The new agencies of the learning machine are also literal-minded. If we program a machine... and ask for victory and do not know what we mean by it, we shall find the ghost knocking at our door.

In a survey carried out in 2016 and 2022 by the AI Impacts project among more than 700 AI experts, the results obtained are those of the figure, obtained from economist.com. In six years, the rather optimistic forecasts of 2016 (45% good, 20% neutral, 15% bad) have become clearly worse: 30% good, 15% neutral, 15% bad. It is curious that the 20% lost by the good and neutral forecasts have not been transferred to the bad forecasts, but (I assume) to those who don’t know or don’t want to answer.

But perhaps the question is deeper. Perhaps someone is trying to take advantage of advances in AI to better control humans. Basically, this is the real problem: not whether we will be controlled by machines (or not), but whether a few human beings will control others. That has always been (and continues being) the goal of every dictator. We must not forget that in today’s world there are many more potential or actual dictators than those everyone knows about.

The same post in Spanish

Thematic Thread about Natural and Artificial Intelligence: Previous Next

Manuel Alfonseca

No comments:

Post a Comment