Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Chatbots: Still dumb after all these years

mindmatters.ai

Post reply on HN
Print view View original

351–360 of 426 posts

Re: Chatbots: Still dumb after all these years

#351

Post by mercutio2 »

solveit wrote: post #283
version_five wrote: post #181

Earlier quoted context omitted.

I've often argued chatbots are much worse than an FAQ. With proper structure, like you say, it can be easy to see what is and isn't covered, and for things that are included, look up the answer. A chatbot is like an inefficient front end on an FAQ, you have to guess what they might be able to do, and guess what might trigger them to understand you. Best case scenario, you get the chatbot to provide you with a "help"…

You are correct for people who can skim a few pages of information, identify what is relevant (possibly spread out across several chunks), and synthesize the information into what they need. For most people on HN, this is as natural as breathing air. For a significant portion of the entire population, this is a pretty daunting task. If you're the kind of person who, like me, is frustrated that everything is a twenty-…

I have never heard this particular juxtaposition (YouTube link?!? Why!!!).

But I'll note that I have extremely effective-at-reading-and-synthesizing friends who enjoy watching/listening to YouTube, so it may not be a complete explanation for the nefarious spread of horribly inefficient videos that should've been text.

Re: Chatbots: Still dumb after all these years

#353

Post by redisman »

avereveard wrote: post #132
pwinnski wrote: post #13

Having worked in ML at two different companies now, I think that people interpreting model output as intelligence or understanding says much more about the people than about the model output. We want it to be true, so we squint and connect dots and it's true. But it isn't. It's math and tricks, and if human intelligence is truly nothing more than math and tricks, then what we have today is a tiny, tiny, tiny fraction…

> We want it to be true Founders and salesman pretend it to be true for that sweet sweet VC money, while underneath devs try to fake it as plausibly as possible. And I'll up it with a prediction: watch closely call centers, and as soon as you see them closing in droves, invest in ai startups as something begun to really move.

I interviewed for a “AI” company and asked them about their tech that could take phone calls and extract orders! Wow really cool and impressive! How do you solve X? Oh we have a call center in Mexico that takes all the calls. So you have no usable AI tech at all? Nope. Ok nice talking to you.

They had signed with a major retail company. They were actually making software for the call center. And the “CTO” was “working on” the “AI” in parallel. The company name had AI in it too

Re: Chatbots: Still dumb after all these years

#354

Post by runarberg »

ShinTakuya wrote: post #332
foolfoolz wrote: post #291

Earlier quoted context omitted.

don’t generalize these things. self driving in every possible scenario yes is probably a while off. self driving in specific circumstances is already here, such as airport shuttles and moving shipping containers in docks. you don’t need to solve the general case to be useful

This. And in fact I think that may be the path forward - just update all the main roads to be self driving friendly, like a train network, and keep the smaller roads as manual driving only.

So why “just update” the highways to operate like a train network, instead of just building a train network? What benefits will autonomous highways bring which normal traditional (or even automated) trains don’t?

If you are suggesting we update all but minor streets, it is actually not that simple, since we have to think about pedestrian crossings, bike lanes, parked delivery vans, emergency vehicles etc. Accommodating dumb self driving cars will have to be at the cost of access for all other modes, and that will never happen outside of highways. And if we are going to be spending a bunch of money on intercity travel infrastructure, a well served train line is always a better choice.

Re: Chatbots: Still dumb after all these years

#355

Post by kirso »

version_five wrote: post #42
pwinnski wrote: post #13

Having worked in ML at two different companies now, I think that people interpreting model output as intelligence or understanding says much more about the people than about the model output. We want it to be true, so we squint and connect dots and it's true. But it isn't. It's math and tricks, and if human intelligence is truly nothing more than math and tricks, then what we have today is a tiny, tiny, tiny fraction…

I think a problem is the tighter cycle between academic discoveries and business people trying to monetize them. Large language models were developed, objectively a great achievement, and immediately someone nontechnical wants to apply their own interpretation and imagine that we can build a chatbot that you won't have to pay, and before you know it, people are selling and even deploying them. Anyone who questions or…

I love this answer because it resonates. As a former business person who learned the technical ins and outs, I believe the reason why this is happening is because majority of the population is uneducated how general tech works under the hood. I am not even speaking about coding syntax but the abstractions of what does AI actually mean, how it works with data and what are the limitations.

The vast majority of business people think they get it, but they majorly overestimate the work required to actually produce output (whether it's ML or software in general). However that's hard to do when you haven't actually done it (talking about deliberate practice).

Despite that gap, we still need to push commercially viable apps out there to seek progress, the question is rather what is the gap between the reality and expectations and what is really being marketed as a capability.

Re: Chatbots: Still dumb after all these years

#356

Post by throwaway14356 »

lotsofpulp wrote: post #208
klabb3 wrote: post #195

Earlier quoted context omitted.

> Anyone who questions or points out that the technology doesn't do what the business people think it does[...] Uh oh, we've got a downer! Jokes aside, I'd like to consider an even simpler explanation, namely that "The purpose of a system is what it does"[1]. In this case, it would suggest decision makers are fully aware that they suck. Why would anyone want something that sucks? Because it's discouraging, and custom…

I have to agree with this. I do not understand why I type all of my relevant information into a chatbot, and then a real person comes on the other end, and asks me all the same questions. Worse is when it is on the phone. I can only assume the intent is to discourage me, as that amount of ineptness is even more depressing to assume.

I call a company one time and the phone was answered in 1/2 second, the person picking up clearly got all the details on their screen as she correctly guessed why i called (which was not all that obvious) then suggested a good solution and (now 2 sec into the call) told me to call again if i had more questions, to have a nice day and ended the call. It went so fast im still laughing how right they got it. Its a great experience, it feels like the helpdesk is with you all day, always ready to answer instantly. It doesnt matter if their competition is cheaper. I would recomend them to anyone over anything.

Re: Chatbots: Still dumb after all these years

#357

Post by wildrhythms »

AnIdiotOnTheNet wrote: post #131
arthur_sav wrote: post #5

I really hate when companies hide behind chatbots. Like, NOOOO, your "AI" chatbot didn't help me - get me an actual person.

Honestly I've had plenty of real world support people who were just as bad, if not worse, than the AI bots. Recently had an experience with paid Microsoft support (for work) so bad that we just stopped even talking to them. It didn't used to be this way, it used to be that if we had a data corruption issue with SQL we'd talk to an engineer who worked on SQL server at Microsoft, now we talk to some third party company…

And this is why chat bots are very attractive to the executives who view support technicians as humans who exist simply apologize to a customer and follow a script.

Re: Chatbots: Still dumb after all these years

#358

Post by redisman »

scotty79 wrote: post #193

Just try to use GPT3 as a chatbot. It will blow your mind. Sure, it thinks it's 2017 and knows nothing about covid and if you ask it to explain math to you it will serve you pure garbage in most educational tone and manner, but it keeps context of the conversation orders of magnitude better than anything else. It doesn't feel like a bot. More like a dumb and confident human.

I did try it. It did not blow my mind. The experience is not really any different from the state machines of olden times unless you’re only asking easy to Google questions

Re: Chatbots: Still dumb after all these years

#359

Post by asdff »

ShinTakuya wrote: post #332
foolfoolz wrote: post #291

Earlier quoted context omitted.

don’t generalize these things. self driving in every possible scenario yes is probably a while off. self driving in specific circumstances is already here, such as airport shuttles and moving shipping containers in docks. you don’t need to solve the general case to be useful

This. And in fact I think that may be the path forward - just update all the main roads to be self driving friendly, like a train network, and keep the smaller roads as manual driving only.

It really does seem like the self driving problem becomes so much easier when you remove variables. Something like an HOV lane would be perfect for self driving. You could put sensors under the road bed to stay in lane even without visual feedback. You could limit it to self driving vehicles pretty easily too compared to a surface street, where you have to account for probably an order of magnitude more variance in conditions.

Re: Chatbots: Still dumb after all these years

#360

Post by lotsofpulp »

throwaway14356 wrote: post #356
lotsofpulp wrote: post #208

Earlier quoted context omitted.

I have to agree with this. I do not understand why I type all of my relevant information into a chatbot, and then a real person comes on the other end, and asks me all the same questions. Worse is when it is on the phone. I can only assume the intent is to discourage me, as that amount of ineptness is even more depressing to assume.

I call a company one time and the phone was answered in 1/2 second, the person picking up clearly got all the details on their screen as she correctly guessed why i called (which was not all that obvious) then suggested a good solution and (now 2 sec into the call) told me to call again if i had more questions, to have a nice day and ended the call. It went so fast im still laughing how right they got it. Its a great…

For sure, there are some businesses out there that do do it right and have your info ready to go based on called ID.

But you have 99% of businesses that do not even offer a call back system, including the government. In 2021, if you do not offer call back systems or email, then it can be assumed you are intentionally being hostile to customers who need help.

Post reply on HN