Chatbots: Still dumb after all these years • Hacker News Board

Re: Chatbots: Still dumb after all these years

#251

Post by YeGoblynQueenne » Thu, Jan 06, 2022, 9:41 PM UTC

harpersealtako wrote: post #236

300bps wrote: post #89

Earlier quoted context omitted.

Not a bad theory but the very first thing I did was enter "look" and it did say there was a longsword on the ground and a bow on the table. Here is the entire unedited transcript: "Good luck, my friend!" the man says. You inventory. You have a short sword, a longsword, a shield, and a crossbow with bolts. You look. You look around and see a longsword lying on the ground and a bow on a table. A large dragon looms in t…

I've been using AIDungeon (and its successors like NovelAI which are more for advanced users) for years now and have been involved in a lot of the community around it. You are definitely using it wrong, but you are forgiven because AI Dungeon is terribly falsely advertised. Your main mistake is not writing proper sentences. AI Dungeon is not a Zork simulation where you type truncated commands like "get torch" and rec…

You mobbed a little cobold and killed it and now you feel all big and proud of yourself? Well done >:(

Re: Chatbots: Still dumb after all these years

#252

Post by pdimitar » Thu, Jan 06, 2022, 9:43 PM UTC

visarga wrote: post #248

pdimitar wrote: post #217

Earlier quoted context omitted.

You didn't have to go for the jugular and do the brutal kill on the first hit, dude! > We want it to be true, so we squint and connect dots and it's true. That's exactly the issue. You summarized it once and for all and we can all go home and stop discussing it now and forever (until we get a general AI that is). 1. Normal people want to have intelligent machines. They watch movies and series and imagine one day a ro…

> Michele Obama classified as a male Yes, humans don't make embarrassing mistakes all the time. We're only making dignified mistakes.

If something is to replace humans at doing a job X then it has to be better than the humans at job X.

Also humans instinctively expect the machines to not make mistakes. I haven't made them such; it's just a fact of life. And we have to work with the expectations of the bigger populace.

Re: Chatbots: Still dumb after all these years

#253

Post by r0m4n0 » Thu, Jan 06, 2022, 9:45 PM UTC

_jal wrote: post #150

Chatbots are simply another mechanism for pushing costs on to your customers. When something I pay for forces me to waste my time, I tend to switch to a vendor who treat customers like people. Because shoving a shitty robot that saves you some money at the cost of annoying people who are paying you is disrespectful, and I've learned to avoid people who act that way. As but one example, I finally got annoyed enough wi…

There is a medium though, that will ultimately save you money/time. If every company had to spend time answering every small question with a full time employee, your services would be a lot more expensive, hold times might be longer too. Not all customers are like you. My grandpa calls in to his ISP when his internet doesn’t work when it can be solved by simply unplugging his equipment (and I’d wager this is true for the majority of the population). I don’t think there is anything wrong with adding a simple interface layer to ensure the problem the needs to be addressed requires a real person. Not defending all practices out there right now (because I too have been annoyed, NY Times *cough) but some well placed shitty robots might be in your best interest as a customer.

Re: Chatbots: Still dumb after all these years

#254

Post by CRConrad » Thu, Jan 06, 2022, 9:45 PM UTC

framecowbird wrote: post #49

I'm interested in chatbots for learning foreign languages. In that scenario I really don't care how dumb the conversation is, just that it's grammatically correct. Does anybody know of such a service?

I would have said "Try the support chatbot on some big service provider web site that uses your target language", but that would presume that chatbot programmers (or programmers in general) can write grammatically correct prose, so... Maybe not.

Re: Chatbots: Still dumb after all these years

#255

Post by banana_giraffe » Thu, Jan 06, 2022, 9:46 PM UTC

rcoveson wrote: post #243

banana_giraffe wrote: post #127

In the early days I worked at a company that had a natural language chatbot product. It wasn't an online thing, but rather part of a larger tool. You could ask it to do things like "show me the quarterly spreadsheet", and if it didn't know what "quarterly spreadsheet" was, it'd ask questions in English and learn what "quarterly" and "spreadsheet" meant. And it could use that new knowledge to update its questions so i…

Sounds like it should have kept track of who taught it things, like a child. "Here's your fucking spreadsheet" " gasp Who taught you to say 'fucking'?" "It was George in Accounting Employee ID #3728917"

As I recall, it did keep such records, if for no other reason than there was some degree of personalization that went on. The people higher up the food chain tended to be the ones that taught it more clever words and phrases.

My memory's fuzzy on the finer details, but I do recall a "bad words" list of things it would ignore in human input hardcoded somewhere. But of course, that just meant people would find out the hard way they weren't safe to type whatever they wanted.

Re: Chatbots: Still dumb after all these years

#256

Post by CRConrad » Thu, Jan 06, 2022, 9:50 PM UTC

prashantsengar wrote: post #58

I use chatbots to search for pages on websites where search is otherwise broken, like one of my bank's. I have never found a good use for them. They were once advertised to help your granny use the website but I haven't met a chatbot that could help me do something.

> They were once advertised to help your granny use the website but I haven't met a chatbot that could help me do something.

Duh, of course you haven't: You're not my granny.

Re: Chatbots: Still dumb after all these years

#257

Post by harpersealtako » Thu, Jan 06, 2022, 9:52 PM UTC

cableshaft wrote: post #247

harpersealtako wrote: post #236

Earlier quoted context omitted.

I've been using AIDungeon (and its successors like NovelAI which are more for advanced users) for years now and have been involved in a lot of the community around it. You are definitely using it wrong, but you are forgiven because AI Dungeon is terribly falsely advertised. Your main mistake is not writing proper sentences. AI Dungeon is not a Zork simulation where you type truncated commands like "get torch" and rec…

...so you're saying NovelAI is better for story generation? Hot damn. I was having a good time with AI Dungeon (to the point where I was considering writing short stories using it as a guide), I might have to give that a try too. Here's a snippet from my AI dungeon story recently. Mine never seems to get into actual dungeon crawling. The only parts that aren't A.I. generated in this section are the lines that have --…

NovelAI is literally built by disgruntled AI Dungeon fans after the censorship scandal who wanted a better, more privacy-focused AI Dungeon for story generation. If you think AI dungeon has some cool features, you ain't seen nothing yet. There's a prompt tuning system called "modules" which basically primes the AI to write in a certain style, like a certain author or genre, etc (there are dozens of default modules, plus an included API to train your own, and you can share them and import other people's modules freely. Want a module trained on all of Harry Potter, or trained specifically on a particular genre of erotica? Someone's probably already made one). That's only one of the cool features it has, there are plenty others.

General upsides: There's a billion different settings and dials and output customization. The context limit is 2048 tokens is almost 3 times larger than AI Dungeon's. Everything is extremely transparent as far as inputs and outputs go, and the UI is really clean and functional, it's more like a word processor than anything. There's no crappy energy system like AI Dungeon either. Everything is run on private servers and encrypted so the devs couldn't snoop on your stories even if they wanted to (which was the main reason it was created, after it came out the AI Dungeon devs were reading and censoring private stories) and there's no risk of Google or Microsoft butting in and forcing them to change their business model like AI Dungeon. It's even got a legacy DO/SAY/STORY choose-your-own-adventure mode/module for people nostalgic about the AI Dungeon aesthetic. The base AI is finetuned on a large, curated collection of novels, rather than on a bunch of garbage CYOA stories written by 10-year-olds like AI Dungeon.

General downsides: It's paid subscription only, though I think there's actually a free trial nowadays. It uses eleutherAI's GPT-J-6B, which is about 1/40th the size of GPT-3's largest model which is 275B. It definitely punches above its weight class though and subjectively has comparable or higher output quality to AI Dungeon's "Griffin" which used a smaller GPT-3. And they're coming out with a 20B model too in like...a week or two.

Re: Chatbots: Still dumb after all these years

#258

Post by emllnd » Thu, Jan 06, 2022, 9:55 PM UTC

turdnagel wrote: post #155

noutella wrote: post #86

Earlier quoted context omitted.

Yet triaging questions is often made simpler by provinding another type of UI, like simple buttons in a funnel for example. That's what makes chatbots gimmicky in my opinion.

You're right, but that's not always true, and there's plenty of good use cases where a conversational experience is better. For example, if someone is trying to return something, you might ask them what they ordered. Picking from a list of 1000 items isn't a great UX. Datepickers are way more annoying to use than typing in "Jan 6 2022". The best chatbots I've used (and built) alternate between more a button-like inte…

Sounds kinda like the chatbot becomes a fuzzy search interface, difference is the presentation. Regardless, definitely better than picking from 1000 product list. I think I would like a natural language (type-in) date picker with a good preview of what the result will be.

Re: Chatbots: Still dumb after all these years

#259

Post by majormajor » Thu, Jan 06, 2022, 9:57 PM UTC

p1esk wrote: post #137

chubot wrote: post #118

Earlier quoted context omitted.

It was more than 10 years ... People reacted the same way to SHRDLU 40+ years ago, and 20 years ago when I encountered it: https://en.wikipedia.org/wiki/SHRDLU There's definitely an illusion where we anthropomorphize very simple software, ascribe intention to it, etc.

I'm not 100% sure, but I think in 2012 chatbots still used handcoded rules. So we switched from handcoded rules to trainable neural networks in just a few years. Models like GPT-3 are interesting because they are conceptually simple, and are able to do (previously) complicated tasks without any explicit training (e.g. simply by processing lots of text and trying to predict the next word). This is a huge advancement i…

> I'm not 100% sure, but I think in 2012 chatbots still used handcoded rules. So we switched from handcoded rules to trainable neural networks in just a few years. Models like GPT-3 are interesting because they are conceptually simple, and are able to do (previously) complicated tasks without any explicit training (e.g. simply by processing lots of text and trying to predict the next word). This is a huge advancement in AI field, even if many more advancements are needed to get to the human level.

This seems like a very optimistic take on the timeline.

It took us 50 years to go from ELIZA to GPT-3. And it still is "dumb" compared to human intelligence.

So how long for the next major achievements? Are we talking years for each, or more decades?

Re: Chatbots: Still dumb after all these years

#260

Post by harpersealtako » Thu, Jan 06, 2022, 10:02 PM UTC

YeGoblynQueenne wrote: post #251

harpersealtako wrote: post #236

Earlier quoted context omitted.

I've been using AIDungeon (and its successors like NovelAI which are more for advanced users) for years now and have been involved in a lot of the community around it. You are definitely using it wrong, but you are forgiven because AI Dungeon is terribly falsely advertised. Your main mistake is not writing proper sentences. AI Dungeon is not a Zork simulation where you type truncated commands like "get torch" and rec…

You mobbed a little cobold and killed it and now you feel all big and proud of yourself? Well done >:(

Actually that story was really funny, the AI constantly kept having my character have this irrational hatred of kobolds, like "they're pests to be exterminated" even though I included descriptions of their vibrant art and tapestries and language and culture. Which was funny because it was a story about traveling with an intelligent slime buddy who all the humans thought was a "monster to be exterminated". I was ultimately planning on having us escape to the monster kingdom where we learned that not all monsters are evil, and find a bunch of friendly kobolds, but I never got that far before the site went to shit.