zlacker

[parent] [thread] 13 comments
1. mrweas+(OP)[view] [source] 2023-12-18 13:27:52
Can someone who understand LLMs and ChatGPT explain how they expected this to work? It looks like they just had a direct ChatGPT prompt embedded in their site, but what was that suppose to do exactly?

I can understand having an LLM trained on previous inquiries made via email, chat or transcribed phone calls, but a general LLM like ChatGPT, how is that going to be able to answer customers questions? The information ChatGPT has, specific to Chevrolet of Watsonville can't be anymore than what is already publicly available, so if customers can't find it, then maybe design a better website?

replies(5): >>mrtksn+12 >>gnz11+u3 >>chasd0+U4 >>paxys+zc >>wharvl+Yt
2. mrtksn+12[view] [source] 2023-12-18 13:35:45
>>mrweas+(OP)
The OpenAI platform can utilize function calling and documents(you can upload files which ChatGPT can refer to). For examples, you can build an assistant that knows specifics about your product and can take actions for you, it can offer the customer a car from the inventory with the requirements they demand and schedule a test drive appointment. You don’t have to engineer or train an LLM, you can simply tell an existing one to act in a specific way.

In this particular case they screwed up the implementation.

replies(1): >>wahnfr+aj
3. gnz11+u3[view] [source] 2023-12-18 13:42:42
>>mrweas+(OP)
What’s there to explain? Contractor company that built the website up sold the dealer on AI chat bots. Contractor company slapped some nonsense together, sold it to naive dealerships who just said “yup, sounds good.” Some irony in a car dealership getting fleeced like that.
4. chasd0+U4[view] [source] 2023-12-18 13:48:59
>>mrweas+(OP)
> It looks like they just had a direct ChatGPT prompt embedded in their site, but what was that suppose to do exactly?

Every actual application of an LLM in prod that I’ve seen has only been this. A better self service or support chatbot. So far, not exactly the “revolution” being advertised.

5. paxys+zc[view] [source] 2023-12-18 14:21:26
>>mrweas+(OP)
"I need an SUV for my family of 5. Which one should I buy?"

"What is the gas mileage of the Chevy Colorado?"

"What electric vehicles are in your lineup?"

"What is the difference between the Sport and Performance models of the Equinox?"

Feed the LLM the latest spec sheet as context and give it a few instructions ("act as a Chevy sales rep", "only recommend Chevy brand vehicles", "be very biased in favor of Chevy...") it can easily answer the majority of general inquiries from customers, probably more intelligently than most dealers or salespeople.

replies(3): >>hacker+Af >>throwa+4l >>deely3+VH2
◧◩
6. hacker+Af[view] [source] [discussion] 2023-12-18 14:36:51
>>paxys+zc
that "easily" is carrying a lot of weight. notwithstanding how AI is simply vulnerable to SQL injection / CB's example / etc, except unbounded through natural language
replies(1): >>paxys+Wg
◧◩◪
7. paxys+Wg[view] [source] [discussion] 2023-12-18 14:43:15
>>hacker+Af
Sure it is vulnerable to prompt injection, but the only one affected by it is the person doing the prompting. Outside of "haha look I made it say a funny thing" there is really no side effect and no disruption for regular users of the service.
◧◩
8. wahnfr+aj[view] [source] [discussion] 2023-12-18 14:54:47
>>mrtksn+12
If this is a screw-up, what isn’t? You’re saying it’s user error rather than the tech being ineffective, so what sales chat bots are correct?
replies(1): >>mrtksn+2n
◧◩
9. throwa+4l[view] [source] [discussion] 2023-12-18 15:04:31
>>paxys+zc
This is a great reply. People here are overestimating how much intelligence (rational thinking) that people put into buying a car. For most people, it is about sales / emotions. If ChatGPT can help to sway a buyer, it is a win for the dealership.
◧◩◪
10. mrtksn+2n[view] [source] [discussion] 2023-12-18 15:14:27
>>wahnfr+aj
I don’t know other sales chat bots, I’m simply explaining how this works. It appears that they improved the implementation later.

Besides, what makes you think that it’s ineffective? Any reason to believe that the chat bot was bad in fulfilling legitimate user requests? FYI, someone making it act outside of its intended purpose affects only that person’s experience.

It’s a DAN attack, people are having lots of fun with this type of prompt engineering.

It’s just some fun in the expense of the company paying for the API. The kind of fun that kids in the early days of the web were having by hacking websites to make it say something funny - just less harmful because no one else sees it.

11. wharvl+Yt[view] [source] 2023-12-18 15:42:02
>>mrweas+(OP)
Owner/exec/whatever: reads some bullshit about AI

“OMG you guys, we can save so much money! I can’t wait to fire a bunch of people! Quick, drop everything and (run an expensive experiment with this | retool our entire data org for it(!) | throw a cartoon bag of cash at some shady company promising us anything we ask for)! OMG, I’m so excited for this I think I’ll just start the layoffs now, because how can it fail?”

- - - - -

The above is happening all over the place right now, and has been for some months. I’m paraphrasing for effect and conciseness, but not being unfair. I’ve seen a couple of these up-close already, and I’m not even trying to find them, nor in segments of the industry most likely to encounter them.

It’d be very funny if it weren’t screwing up a bunch of folks’ lives.

[edit] oh and for bigger orgs there’s a real “we can’t be left behind!” fear driving it. For VC ones, they’re desperate to put “AI” in their decks for further rounds or acquisition talks. It’s wild, and very little of it has anything to do with producing real value. It’s often harming productivity. It’s all some very Dr Strangelove sort of stuff.

replies(2): >>soupfo+271 >>Corrad+o02
◧◩
12. soupfo+271[view] [source] [discussion] 2023-12-18 18:38:58
>>wharvl+Yt
Lived this as well. Even more painful when you actually try to explain it to them.
◧◩
13. Corrad+o02[view] [source] [discussion] 2023-12-18 23:25:25
>>wharvl+Yt
I just got back from the AWS re:Invent conference and it was full of AI stuff, most of which didn't make much sense. The biggest announcement was "Amazon Q" [0], the Amazon general purpose chatbot. They hooked it up to the AWS console and I've not found a single reason to use it. I tried a couple of questions about a problem that I was having and it didn't provide even a modicum of help. So far, I see GAI as a complete failure.

[0] https://aws.amazon.com/q/

◧◩
14. deely3+VH2[view] [source] [discussion] 2023-12-19 06:27:31
>>paxys+zc
And after each question please go and do a search manually on the web to verify answer.
[go to top]