[mmooss]

Why allow anthropomorphism in LLM or any computer output, for any age group?

A simple regulation would be that all LLM output must:

  * Explicitly identify itself as a computer

  * Identify the human or human org controlling it
    (e.g., 'This is Claude v8.90 software made by Anthropic.')

  * Not attempt to mimic humans except for entertainment
    purposes, and then clearly labeled (e.g., 'This is not
    a human. It is a computer pretending to be human for
    entertainment purposes.').

What violation of those regulations would be for other than defrauding and/or manipulationg people? They are easily complied with and I don't see those regulations impacting the economic productivity gains projected for LLM use.

[dpark]

This sounds like the kind of well-intentioned thinking that results in products that are no safer practically but are far more unpleasant to interact with. This is like the pop up every time I start my car that reminds me I should not play with the screen while I’m driving.

[mmooss]

Would you give examples of possible negative results, especially those that outweigh the benefits? Do you have ideas on how to do it better?

> like the pop up every time I start my car that reminds me I should not play with the screen while I’m driving.

If something like that is the worst result, there is little downside.

[dpark]

The downside is it’s annoying as hell. The upside seems limited to a bureaucratic pat on the back for “doing something good”.

[mmooss]

I don't see any basis offered for what you're saying.

[dpark]

The annoyance of every AI announcing is a machine every time it’s invoked is obvious. I don’t see any basis offered for this actually doing anything beneficial.

[victorbjorklund]

What does it even mean ”do not mimick humans”? How are the AI supposed to reply if I ask ”do you think this genserver will be a bottleneck in the sync process and how can we fix it?”

The correct answer would sound just like a human unless you either make it do shallow annoying stuff like add redundant ”as an LLM I can’t think or recommend but the answer with the highest probability is …”

[xg15]

More "Star Trek computer" and less C3PO would be a start.

We all know the "voice" of log output of normal, non-LLM computer programs. Usually, the statements are in passive voice and tersely describe what the program is doing or print new kind of data available. An LLM with that style would already sound much more like a computer program than a person.

E.g. an answer to your example prompt could be something like:

   * Locating relevant code for "genserver" and "sync" in the codebase...
   * Reading files...
   * Data collection completed. Analysis results are as follows:
     * Factors that suggest a bottleneck are:
        - ...
     * Factors that speak against a bottleneck are:
        - ...
     * Overall assessment:
        - ...
   * Request complete.

Or something like that.

[victorbjorklund]

Sounds just like bad UX and harder to use.

[mmooss]

What about it is 'bad UX' or harder? It seems simple, effective, and more efficient to me.

[mmooss]

Computers long have always, before LLMs, and most still do do output responses that don't sound like humans. It's hard to understand why you think that's hard.

It is a matter of definition, but a start would be eliminating the social and conversational aspects and attempts at emotional connection.

[marcus_holmes]

Isn't this about children attributing human characteristics onto the LLM, rather than the LLM itself pretending to be human?

And we do this with lots of things that are not human for children (e.g. Bluey and Peppa Pig). Do we need a "this is not a human" warning for them, too?

[estearum]

Do Bluey and Peppa Pig produce convincing human-like responses to virtually any human-generated input directed towards them?

If so, then uhh yeah we should definitely regulate that shit

[marcus_holmes]

If not now, then soon, yes. I'll bet money there's a toy manufacturer somewhere working out how to hook a Bluey doll up to an LLM to make it responsive.

I'm not completely clear on what we're afraid of here, though. We routinely project human-like characteristics onto non-human things for our children. We don't think this is harmful. Why are we worried about children projecting human characteristics onto an LLM?

[estearum]

Obviously the people who think LLMs should be forbidden from this use-case would think it should be forbidden from put-it-in-a-doll use-case.

We routinely give our children ibuprofen or acetaminophen, this does not imply that some vicodin or oxycontin would be fine.

I think it's a reasonable assumption that a fundamental skill of being a good social creature (which we know to be absolutely critical to good human health) is to be able to distinguish between human and non-human. Or more generally conscious and non-conscious.

You are basing your opinion off of e.g. anthropomorphizing cars or dolls which 1) are obviously not human, and 2) rely on the fairly intense imagination of the anthropomorphizer to "work." Once the imagination stops, the anthropomorphization stops.

In contrast, LLMs successfully delude even intelligent adults into anthropomorphizing them. This is especially problematic for a generation of humans that grow up where most of their human-to-human communication happens intermediated by technology, i.e. LLM-originated interaction looks more human every day, but also human-originated interaction looks more LLM- (or technology-)originated every day too.

We've been living with this pathology for a while now in the form of people treating those humans on the other side of a chatbox vastly worse than they would ever treat a human face to face. Now this is chewing away at the distinction from the other end too.

[marcus_holmes]

I'm always wary of extrapolating from the few individuals who experience that harmful interaction with chatbots to the general population. We have the same thing with every new tech, that some people react badly/too strongly to it and get into a mental health situation [0]. There's nothing unusual about LLM chatbots in this respect. So yes, some people anthropomorphise LLMs too much, but that's not a statement you can make about the general population as "even intelligent adults anthropomorphise LLMs too much".

We are also talking about kids, who will grow up with this technology around them and adapt to it better than we do as adults. We see it as a threat and a change to the natural order of things, because we grew up without it. Kids growing up with it will see it as part of the natural order of things and will learn how to incorporate it into their lives. That will take some adaptation, sure, but that's the same with every new technology. The answer is not to shield them from it, because we just put off the day that they get exposed to it with no training. We must coach and teach them how to deal with it healthily. Of course, that means that we ourselves as adults need to work out how to deal with it healthily first, which isn't always obvious ;)

[0] https://en.wikipedia.org/wiki/Internet_addiction_disorder

[estearum]

Does this techno-optimist logic apply to all technologies, e.g. glyphosate or oxycontin, or only to technologies that you first presume are safe/positive, and then you can just argue from that presumption?

[smcg]

We probably do, considering how many seem to treat it as companionship or therapy.

[marcus_holmes]

Yeah that's adults, though. Unless we go down the McDonalds coffee warning path "warning: this LLM is not human and can harm you if you treat it as human"

[ndnichols]

Preach! Mustafa Suleyman had a great post on some of this last year: https://mustafa-suleyman.ai/seemingly-conscious-ai-is-coming

[pixl97]

Because it makes the product mostly useless for all kinds of things?

[windexh8er]

Care to enlighten us? Because you seem to imply you want to have a conversation about your day every time you fire up a shell.

[mmooss]

What things? Again, I don't see any but manipulation and fraud, but I certainly could be missing some.

[all2]

Role play, which is a huge consumer of tokens. You can glance at openrouter's usage stats and see what apps tokens are going to and what the inference content was classified as. It's quite revealing.

[mmooss]

How would that regulation interfere with role play? Role play is 'play'; people know what it is.

It shouldn't interfere with immersion: People suspend disbelief in movie theaters or reading books, despite knowing it's a film or novel. The risk with AI is that people don't understand what an LLM is.

[estearum]

Yeah, the costs of that use case far outweigh the benefits.

[selectodude]

Like online gambling, some things cause more harm than good.

[BrenBarn]

I'd say making LLMs useless would be a net positive for humanity.

[4b11b4]

What world are you living in, I don't want to go there

[ericol]

I don't need a computer to mimic a human. AT ALL

As a matter of fact, for the line of work I do (programming, devops, the works) being conscious about the fact that I am just getting an algorithm outputting text that is probabilistically related to the input I gave it would actually be healthy for me and my mental state.

I HATE having to say "you" when working with an LLM.

[mmooss]

Don't say 'you'?

One solution is to use commands (imperatives), as you've always done with computers. For example, instead of 'you are an expert in Lisp', say 'function as an expert in Lisp.'

Are there use cases where that isn't applicable.

[ericol]

I don't think that works at all times; actually I think it only works when starting a conversation. In the middle, "Why ~did you~ while action as an export in Lisp <this happened>?" needs a certain level of mind gynmastics that is not worth it. I'd rather curse at it.

[mmooss]

I do wonder if there are circumstances where it doesn't work. But it's hard to believe I'm addressing basic syntax for computer commands on HN:

'Display the cause of the <this prior output>'.

> needs a certain level of mind gynmastics that is not worth it. I'd rather curse at it.

I guess I knew an LLM's natural language interface made it easier to use for non-technical users. But wow, it's been normalized so quickly. Now the universal old way, needed for every line of code and command line and more, is "mind gynmastics"?

Inputting curses seems just odd to me, beyond the initial novelty. It's a distraction, completely unproductive, and sending emotion into a void.

Anthropomorphisizing it seems dangerous personally and society-wide.

[marcus_holmes]

I have the same urge - third-person the LLM is an it, for sure, but I'm not aware of a suitable second-person construction for talking to things. Maybe we need to invent one.

[wpm]

When yelling at the other appliances I own, "you" usually turns into "you fucker" or "you useless piece of shit", so I stick with what I know.

"Claude, you useless piece of shit, I already told you..."

[marcus_holmes]

Always loved the tweet: "I don't know what machine they were raging against, but I'm pretty sure it was a printer" [paraphrased]. We could definitely update that.

I'm Australian, so "oi cunt" works about as well as "you" for most things, though I'm not suggesting we adopt that formally.

[ericol]

ok that cracked me. Could be worth of an update:

    I don't know what machine they were raging against, but I'm pretty sure it was an LLM

[4b11b4]

Seriously: for the sake of all future children, the elderly, and everyone in between