-3

In 2022, a Scientific American article quoted Google engineer, Blake Lemoine, who argued that LaMDA (Language Model for Dialogue Applications) might be sentient, while other technical experts criticized the statements.

In [a Washington Post] article, Lemoine recounts many dialogues he had with LaMDA in which the two talked about various topics, ranging from technical to philosophical issues. These led him to ask if the software program is sentient.

[...]

Many technical experts in the AI field have criticized Lemoine’s statements and questioned their scientific correctness.

Is LaMDA sentient?

11
  • 6
    I don't think there is anything to add past the Wikipedia page on the controversy.
    – Oddthinking
    Commented Oct 31, 2023 at 5:29
  • 3
    This seems to be just one software developer thinking the program he worked on is special, and everyone else thinking he's being sensational or misunderstanding it. Are there any actual quotes from Lemoine saying LaMDA is 'sentient', or is that just the article's summary of his claims?
    – Giter
    Commented Oct 31, 2023 at 14:43
  • 2
    We're going to need a stricter definition of "sentient" to answer this. Commented Oct 31, 2023 at 18:25
  • 2
    @AgentSmith No, but if you have any info on what Lemoine meant by it, that would be a good starting point. If not, I guess pick your favorite dictionary as a starting point. I've heard roughly as many definitions of "sentience" as II've had conversations about sentience. Commented Nov 1, 2023 at 15:13
  • 3
    What does "sentience" even mean? Can you tell the difference whether something (or even someone) is sentient and doesn't just pretend to be? Can you prove I am sentient? Can you prove you are? Can you prove that a rock isn't sentient? That's a question philosophers debate about since the dawn of time, and so far there doesn't seem to be a consensus.
    – Philipp
    Commented Nov 1, 2023 at 15:23

1 Answer 1

6

If you search for "Is LaMDA sentient", you'll get plenty of articles like https://opendatascience.com/is-lamda-really-sentient-no-far-from-it/ and https://link.springer.com/article/10.1007/s00146-022-01559-z talking about it not being sentient. From the latter:

So, where is the sentience? Well, nowhere, it seems. LaMDA does not have any. It does not experience, it does not feel, and it does not have memories. If you ask LaMDA, "are you happy?" it will generate an answer based on what it has been trained to optimise: language use. It is not going to recall a time when it was happy or sad—that winter where it was lonely because the other servers were ignoring it, or that time the training function was upset because it was not being called enough. We can call this loneliness, but in so doing, we rob loneliness of its meaning.

So, it should be clear that LaMDA is not sentient. Anyone who claims it does has a significant burden of proof to bear.

Also good is https://ai.stackexchange.com/a/36252, which mentions how Lemoine is likely falling victim to the ELIZA effect.

It's also worth noting that if LaMDA were indeed sentient (or there was still a possibility it could be considered as such), we could expect to see some continuing discussion about it. Instead, in the above search I did not see any results from 2023. This is absolutely not solid evidence that LaMDA is not sentient, but it is worth noting.

1
  • Most interesting.
    – Hudjefa
    Commented Nov 1, 2023 at 20:57

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .