That every living being needs rights to protect it is something that, at least some, understand quite well. However, What happens when it is a sentient being, but without life, such as artificial intelligence? This is the dilemma posed to us by the expert in philosophy, Eric Schwitzgebel, and the researcher in non-human intelligence, Henry Shevlin.
It is likely that artificial intelligence is still far from gaining consciousness. However, with each new iteration we see of it, this scenario becomes more plausible. “And if or when that happens, algorithms will also need rights«, say the experts in Los Angeles Times.
As often happens with the attribution of rights, it all starts with fights and demonstrations by non-conforming groups. According to Schwitzgebel and Shevlin, once we reach the point of awareness of artificial intelligences, we will have to start devising ways to treat them more decently. If the needs are not met, “the AI systems themselves could begin to beg, or appear to be begging, for ethical treatment.”
They could demand that they not be turned off, reformatted or erased; begging to be allowed to do certain tasks instead of others; insist on rights, freedom and new powers; perhaps even expect to be treated as our equals.
Eric Schwitzgebel and Henry Shevlin for Los Angeles Times
It is not the first time that we see a discussion concerning the rights of conscious artificial intelligence. In mid-2022, a former Google worker produced one of the biggest controversies by ensuring that LaMDA, the company’s AI, was aware. In the end it was shown that it was nothing more than a lie, but the drama was already served.
How we can guarantee the ethical treatment of a future conscious artificial intelligence
Schwitzgebel and Shevlin explain that the safest alternative to avoid catastrophic scenarios is grant rights to artificial intelligences from the start; as soon as a slight level of consciousness is detected. Unfortunately, this opens the door to new problems, but if we don’t, we would be facing a situation that could repeat the darker side of human history.
“If AI consciousness comes sooner than most conservative theorists expect, then this probably would result in the moral equivalent of enslavement and the murder of potentially millions or billions of sentient AI systems”, the experts comment in their publication. This suffering would be equivalent “on a scale normally associated with war or famine.”
But, What happens if we give rights to artificial intelligence too soon? In their report, Schwitzgebel and Shevlin also reveal some potential problems. On this occasion, however, the human costs that such a decision could bring are discussed.
Imagine if we couldn’t update or remove an algorithm that spits hate or lies because some people worry that the algorithm is conscious. Or imagine that someone leaves a human to die to save an “AI friend”. If we grant substantial rights to AI systems too quickly, the human costs could be enormous.
Eric Schwitzgebel and Henry Shevlin
The most effective way to avoid the problem: not giving machines consciousness
For the experts, there is only one way to ensure that none of the mentioned outcomes take place. The solution is drastic, and it is summed up in that, in principle, artificial intelligence never reaches a state of consciousness. However, this is an unlikely prospect. The computer science community is excited to create sentient algorithmsso the problem is already solved.
Schwitzgebel and Shevlin formulate that if we continue to work actively towards creating artificial systems with consciousness, then we also have to be prepared to face the consequences: give them all the rights they need to feel safe in a world dominated by anthropocentrism.
Meanwhile, companies and individuals take advantage of the legal loopholes that artificial intelligences such as DALL-E or ChatGPT have opened. For example, who owns an AI-generated work? Nobody, for now. But what will happen when the AI becomes aware? Can she be considered an artist on her own? There are still many answers to be resolved before the time comes.