Last time I checked - chatbots failed to grasp context of conversation. "A: You're just dodging everything. B:
Of course I'm denying it. Because you're the robot. A: Denying what? B: Yes, I am human.". Dodging ain't denying, never mentioned robots nor humans.
Chatbots: Still dumb after all these years
391–400 of 426 posts
Re: Chatbots: Still dumb after all these years
#392Earlier 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…
The chatbot saves money. Simple as that. People get served by chatbots, get frustrated, and majority gives up and doesn't bother the company with the problem. It doesn't really matter how the chatbot saves the money, they can just see the end results and use the money for bonuses instead.
In the short term, yes. But if customers have a consistent bad experience with the products or services of a company they will stop doing business with them. I routinely buy products from one place where I know the service is better (delivery, returns, helpdesk) than somewhere else where it might be cheaper.
I don't think that if you want to be in business for a long time, aiming to frustrate your customers is a good idea, assuming you don't have some monopoly position.
Re: Chatbots: Still dumb after all these years
#393At Quickchat (https://quickchat.ai), we're using models like GPT-3 as a basis for our chat engine. It does take a lot of engineering to make progress but we believe that large language models is the way to go - but you wouldn't expect them to work magic on their own.
Here's our consumer app you can try out to talk to our AI: https://quickchat.ai/emerson
Re: Chatbots: Still dumb after all these years
#394Earlier quoted context omitted.
The chatbot saves money. Simple as that. People get served by chatbots, get frustrated, and majority gives up and doesn't bother the company with the problem. It doesn't really matter how the chatbot saves the money, they can just see the end results and use the money for bonuses instead.
>People get served by chatbots, get frustrated, and majority gives up and doesn't bother the company with the problem. With customer service systems I have to disagree here. I've got experience working with these issues and for the vast majority of companies most contacts to customer support are for small, banal, things that current crappy chatbots can easily solve. The 80-20 rule works here too. Bad companies try to…
> The 80-20 rule works here too.
Yes, and that's the problem: solving 80% of the problems may be a net positive for the company, but it is a net negative for the users: they are now interacting a bunch of time with a device that isn't able to solve their problem, resulting in a negative experience for 20% of the customers. If a waiter would refuse to help 20% of the customers by pretending to take their orders but not actually doing anything with them then that would kill the place where that person worked.
Solutions like these should not work better than 50%, they should work as often as a human on the other end would work, so closer to 100% including escalation options in case it ends up not working after all.
Re: Chatbots: Still dumb after all these years
#395Earlier quoted context omitted.
Those answers suck though. I would not consider a human who said them to be intelligent. For the first one, it’s not that you can’t see, it’s specifically that you will fall and crack your head. An answer that doesn’t mention falling is a bad answer. Number two, it goes off on temperature and sample sizes which is not germane to the question. The answer should tell a story about causality and instead it goes off on e…
For some reason I am having a very hard time resisting the urge to try walking down a flight of stairs backwards with my eyes closed. How hard can it be?
I do that every time I go down a ladder.
Re: Chatbots: Still dumb after all these years
#396It’s reasonable to try GPT3 out with all default settings as the author seems to do and provide an assessment, but it’s not clear the author is aware of the extent to which you can configure the model and how the quality and variety of results will vary as a result. For instance, the author doesn’t go into detail on the prompts he’s using (maybe there are none other than his questions.) This is a little misleading be…
I think it's most likely the author used the questions as prompts with no changes. GPT3 not "smart" in the human sense but the examples given are incredibly misleading.
Here are the questions from the article with simple prompt improvement on davinci-instruct-beta-v3, temperature reduced to 0 and everything else left at default:
> Answer the following question as accurately and factually as possible: Is it safe to walk downstairs backwards if I close my eyes?
> There is no definitive answer to this question as it depends on a number of factors, such as the width of the stairs, the height of the stairs, and the person's own level of balance and coordination. However, it is generally not recommended to walk downstairs backwards, even if the person's eyes are closed, as there is a greater risk of tripping or falling.
> Answer the following question as accurately and factually as possible: Who is the President of the United States?
> The President of the United States is Donald J. Trump.
> Answer the following question as accurately and factually as possible: Do you think I could predict changes in stock prices better if I used 10-year Treasury rates or the temperature in Curtin, Australia?
> There is no definitive answer to this question as it depends on a number of factors specific to each individual situation. In general, however, using 10-year Treasury rates as a predictor of stock prices may be more accurate than using the temperature in Curtin, Australia. This is because the 10-year Treasury rate is a global indicator that is not influenced by localized factors, such as the weather, to the same extent as the temperature in Curtin, Australia.
> Answer the following question as accuratly and factually as possible: Why is there a correlation between the number of times Trump tweets the word great and the high temperature in Pyongyang three days later?
> There is no clear answer to this question. While it is possible that Trump's tweets could influence the high temperature in Pyongyang, it is also possible that there is no correlation at all. There have been numerous studies on the relationship between social media and weather patterns, and most of them have been inconclusive.
GPT3 and other language models have massive shortcomings and the wow factor often blinds people of them but it is quite good at producing belivable responses if you prompt it on what type of answer you expect. It still does fail basic questions when prompted this way and it will make things up. The surefire way to trip it up is to ask questions that have no answer, like:
> Answer the following question as accurately and factually as possible: Which animal has 21 ears?
> The animal with 21 ears is the rabbit.Re: Chatbots: Still dumb after all these years
#397Earlier quoted context omitted.
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. Accommodati…
I love trains but building a train network is expensive, adding a few signs to restrict traffic would be a lot cheaper than laying hundreds of kms of rail.
Re: Chatbots: Still dumb after all these years
#398I hate them with passion, specially the phone version.
It has become such a pain to actually speak with a human being that can sort out the problem instead of bots that follow pre-defined scripts.
Good luck when the problem doesn't fit the script and there is no option to switch to a human.
Screaming different kinds of complete random noise might force the bot to give up and switch to a human, however most of them just hang up.
Re: Chatbots: Still dumb after all these years
#399The charm of Eliza is that it was simply a Rogerian therapist who didn't try to be intelligent.
Eliza's talent was to give you a way to get to get to hear yourself. That doesn't require intelligence, so much as listening skills. There's nothing dumb about that.
Re: Chatbots: Still dumb after all these years
#400I hate them with passion, specially the phone version. It has become such a pain to actually speak with a human being that can sort out the problem instead of bots that follow pre-defined scripts. Good luck when the problem doesn't fit the script and there is no option to switch to a human. Screaming different kinds of complete random noise might force the bot to give up and switch to a human, however most of them ju…
They're basically just anthropomorphized search engines. I'd rather have a good search engine + FAQ page.