I recently had the displeasure of buying a car from Carvana. I will explain why in a separate article, but along the way I found something that struck a nerve besides their incompetent/fraudulent sales practices. I have been cautioning people since the emergence of “Generative AI” appearance to not trust it and take what it gives you with a grain of salt. Many companies have blindly jumped into the Gen-AI pool and there has been a predictable reaction to it. The multitudes of people who don’t know any better have been impressed, they ooo-ed and ahhh-ed at what a marvelous parlor trick these stupid “AI” chatbots can do. Finally, most I have argued with have come to the destined conclusion that this is just glorified auto-complete in most cases. Regardless, companies will do whatever they can to pay less to human workers, even if it means potential lawsuits and loss of business because they can only see one foot under their own noses.
Carvana is a trash company
Carvana, sells used cars. They falsely claim that they do a 150 point inspection which they absolutely do not do. I can prove it after going through the horrific experience I just went through with them. The one thing I will give them kudos for is that the buying experience is incredibly easy and enjoyable. If you don’t plan on asking them for help after buying your lemon vehicle, then it’s a great place to buy a broken car. Otherwise, I strongly advise you don’t buy a car from them ever. Selling cars to them is easy, I have done this twice and didn’t have an issue. I would sell them cars again in the future if I needed to.
The chatbot is well trained
The other thing that they did that actually impressed me is they have a chatbot that is actually well trained! It could answer a number of questions I honestly didn’t think it would be able to. I asked it very specific questions about car registration, titles and other car & sales related questions and it pretty much nailed it 99% of the time. I only had to speak to a human a few times via the same chatbot window for things it wouldn’t be able to handle because it required actual thought and decision making.
Deceptive practices
The chatbot is well trained! The other problem is, that the chatbot is well trained! Please look at the chat I had below with Sebastian the chatbot as it blatantly lies to me about information it absolutely has and refused to give to me until I called it out on its bullshit.
What is the lesson here?
This is a demonstration of the programmer training this AI to lie. I asked the AI multiple times what mailing address I should use to mail a complaint letter and all of a sudden it had amnesia. I was testing it from the start because I already knew the answer because it was written in my contract, but I wanted to double check and cross reference the information with what I thought was a pretty well trained bot. Instead, I was being deceived.
There are a few things in this world I have zero tolerance for and they are:
- Fraud
- Lying
If you ever want to see me lose my cool, that’s how you would do it.
At this point I already had – had enough of Carvana’s bullshit and this really just sealed in the ire for me. No one is going to tell me that this chatbot didn’t know this address. It was deliberately trained to: “oops sowwy I don’t know! Tee hee!” and that’s deception. I don’t want to hear from AI apologists on this. Stop using cute-sy language like, “It’s having hallucinations” because all I am hearing is, “There is a bug in the software.” This is deliberate and just an example of what I have been saying since day one, don’t trust this garbage, especially not from a trash company like Carvana. Be skeptical and don’t take the information at face value because it’s absolutely flawed and will continue to be especially when the programmer is pulling stunts like this.
The base thumbnail art was generated using CoPilot. I added in the Carvana logo manually. Yes it’s a low effort thumbnail.