[kleiba2]

Maybe it's just me, but when I was younger one reason that got me into AI research was the idea that one day we would have these seemingly intelligent entities that we could talk to and interact with just like we do with humans.

Now that we have that, everybody hates it.

Except me. I still like this. There might still be room for improvement, yes, but I think this is a cool attempt to making AI agents part of the team.

I think what could help make this better is investing into giving agents actual personality, instead of them being just sycophantic clones of each other. AI agents as team members should have clear aspects of distinction both in their abilities and behaviors.

[1shooner]

>I think what could help make this better is investing into giving agents actual personality, instead of them being just sycophantic clones of each other. AI agents as team members should have clear aspects of distinction both in their abilities and behaviors.

A language interface doesn't need a personality any more than a visual interface does. For most professional work, source data isn't written in the first person. Ego is an imposed product feature that wastes energy. What would the purpose of diverse personalities be? I want models as tools, not friends.

[Sharlin]

Most people likely have a natural aversion against having "team members" without agency or consciousness. Giving them artificial, acted personalities would be even more horrible. LLMs are tools and trying to elevate them to the same level as actual humans is nuts.

(Yes yes, we can't know whether it "feels like" something to exist as bursts of vast numbers of matrix multiplications, but at least it would be a rather strange existence.)

[fidotron]

> Most people likely have a natural aversion against having "team members" without agency or consciousness. Giving them artificial, acted personalities would be even more horrible.

The problem isn't the team member being artificial or not, it's the obvious fakery of the personalities that is the problem, and that's entirely independent from the artificiality of the team member.

[bonoboTP]

> Now that we have that, everybody hates it.

To me, it seems like a loud online minority, and it's mostly entangled with politics or job security.

Most good programmers I know are using it effectively and like it.

[gwerbin]

Using it effectively is not evidence that someone is happy about it. I use it effectively for all kinds of things and I hate quite a lot of it.

The anthropomorophization and cutesiness of this example really sucks to me. I want my AI to be a cold and emotionless tool for making an income in a dying miserable world.

[moron4hire]

I've been at my current company for just over three years. I manage a small team of developers at a medium-sized consultancy, building internal tooling for our data analysts.

One of my junior developers started the same day I did; we went through on boarding together. He's in his mid 20s, this is his second job post-college, after having an abysmal experience entering the workforce during COVID. He's had some struggles with learning to code but has taken my advice to build side projects for things he is interested in and has really started to take off. He got married last year and just last week bought a house.

One of my developers was a brand new junior, having started her first job post-college just a few months before I started at the company. When I first met her, she apologized to me. She hadn't done anything, she was just that kind of mousy person whose obviously been ground down by everyone around her for her entire life. I started making her deliver our project status updates. I also give her full projects that she works on her own with no input from me, other than whatever advice she herself asks of me. She finished her master's degree in computer science last year and also recently bought a house.

One of my more recent developers came to us from after 10 years of working on-site on military bases. She'd been stuck in a corner, made to maintain large legacy projects on her own, over looked for promotion her entire career. She's one of the best project managers I've ever seen. I have her deliver all of our presentations to the C-suite (and there are a lot of them). She just got engaged and I recently got her promoted for the first time in her career.

One of my former developers came to me from a different part of the company. He came to the company after being a medic on active duty in Afghanistan. His former team couldn't figure out what to do with him and when the contract on which he was working expired, they didn't bother to find a new one for him. He had told his supervisor he wanted to be a programmer, not an analyst, and somehow his supervisor knew I was looking for people. He was an absolutely terrible programmer. I spent multiple sessions a week working on mentoring him. Ultimately, I suggested he find another company and helped him find one that I thought would fit him better, something smaller, not in consulting, that could have more time to work directly with him, and kept him busy with documentation tasks until he could transition with no down time. He still sends me updates on LinkedIn. He also bought a house and is getting married soon. He loves his new job, misses our team, but says he's learning a lot more now.

Our company is one of the worst places to work as a software developer I've ever seen. Not quite the worst in that nobody is yelling at us or threatening us with violence, but our management has absolutely zero understanding of the business of software development. It's incompetent-bad, not evil-bad. The actual projects are ludicrously bad with no direction. My boss listens to me in our one-on-ones with bewildered eyes as I describe the cacophony of dumb ideas from our C-suite stakeholders, folks he has no involvement with because my team is matrixed out to internal work. He mostly just lets me do whatever I want as long as the timesheets get signed on time. I love working here. I also recently bought a house, my first one in my life. Oh yeah, we live in Northern Virginia, BTW, houses ain't cheap here. But I love working here despite all the product idiocy because I don't see my job as building software, I see it as building software developers.

Fuck calling a computer program a "teammate." A computer program is not going to have a brilliant and beautiful life. We only do any of this shit because it's in the service of people.

[cattenwallen]

I'm currently new to northern virginia, a software dev. You sound like a great mentor and I would love to meet you for coffee sometime.

[codegrappler]

While I really had to hold on to the end to get the point of the reply, I appreciated the stories. You sound like a great coach/mentor.

[kleiba2]

Thanks for your great input. But you might have mistaken what I was trying to get at - certainly not that there's no difference between a computer program and a person.

[bonoboTP]

> We only do any of this shit because it's in the service of people.

Where does the money come from that was used for those houses? From you? Oh, that's great, then keep handing out housefuls of money to people in your life that you appreciate. What? It was from someone else? Why did that someone else hand over that money? Because of "service of people"? Did they say so?

[kretaceous]

This was really encouraging to read for some reason, thanks!

[latexr]

> these seemingly intelligent entities that we could talk to and interact with just like we do with humans.

> Now that we have that, everybody hates it.

There are several reasons for it, most of them having to do with looking beyond what’s right in front of your nose. For many people, we don’t “have that”, LLMs are not comparable to interactions with humans. Then there’s all the shit around it such as the further consolidation of power in the hands of a few who really shouldn’t have it, the destruction of livelihoods, the increase in disinformation and propaganda, unhealthy habits and depression, and on and on.

Too many of these conversations focus only on the technology and completely ignore the context in which it exists. That lack of human perspective is a major reason why Hacker News is made fun of and dismissed relentlessly on other places on the internet (called “the orange site”), and everyone here is (wrongly) lumped as being a bunch of out-of-touch selfish nerds.

[maeln]

I get it. I am still in love with the idea of having little bot and helper I can interact with. Realistic NPC, etc ... I think my current distrust and disappointment as many reasons that mostly boils down to imagination vs. reality.

The current generation of LLM takes a lot of resources to make, so of course it was made and pushed by the worst offender of late-stage capitalism. Giving an air of mercantilism, corporatism and general greed and unfairness already oh so common in our time. It's hard to see Claude as a cool entity and not exclusively made for profit, with all that entails. It's also (partly via the training data, partly via the proprietary system prompt and other safety mechanism) give the output a god awful personality. Sycophant, safe, easy language, no personality. And I have given up on trying tweaks to not have them sound like an incredible gigantic corporate-safe-speech dork.

Then there is all the limitation of LLM. We keep calling the current tech "AI", but they are as close to my own definition of AI than randomize decision tree npc where in early 2010 games. They can't do logic. They are a random number generator that are just insanely good at giving you the number that you want (which, I should emphasis, is absolutely crazy). Some techniques help with that, but the tech is fundamentally limited, which often make them frustrating.

[Hilliard_Ohiooo]

> They are a random number generator that are just insanely good at giving you the number that you want...

Well, you definitely have the absolute worst definition and least understanding of what an llm is of anyone I've ever met, that's interesting.

[maeln]

Trying to gauge my understanding of LLM based on an exaggerated description not meant for accuracy but to make a point, is also interesting.

[bonoboTP]

It cannot do logic. It just randomly generated a bunch of counterexamples to several open math conjectures. It's clearly unable to reason. It was all luck. Sure.

[conartist6]

So you really place no value on talking to other people. I guess you must think that's not part of engineering

[kleiba2]

Non sequitur.