It's a shame that the itemsy.com bot lies to you:

'What personal data do you take?'

...'We take your email address and nothing else'.

https://itemsy.com/privacy/ - Personal data may include, but is not limited to: Email address, First name and last name, Phone number, Address, State, Province, ZIP/Postal code, City

Whose responsibility is it when the chatbot lies? Itemsy or QuickChat ?

Section 8.2.3.42. The bot's opinion are its own and don't represent the company's viewpoint. You waive your rights to pursue a legal action for any damage caused directly or indirectly by the bot's advice. You waive all your other rights as well and become the company's servant from the moment you have finished reading this.

Actually, the bot could generate very authentic terms of service: 10,000 pages long, written in the most unreadable dialect of legalese.

It also lied to me:

Hey I'm here to answer your questions about Itemsy

Me: It appears you are using third party tracking after all

I'm sorry, I'm not using any third party tracking.

Me: You may be not, but itemsy is

I'm sorry, I did not understand that. Could you explain what you mean a bit more?

Me: Itemsy is using third party tracking

Itemsy is using Google Analytics to track usage statistics.

Me: And yet you said you are not using third party tracking

We are not using any third party tracking.

Me: You see, that's confusing

I'm sorry, I didn't mean to confuse you.

Me: So you are not using third party tracking but itemsy is?

Itemsy is not using any third party tracking.

Me: Yes it is. It uses google analytics

I'm sorry, I'm not sure what you mean. Shall we chat about Itemsy a bit more?

Me: No

I'm a bit confused. Could you say that again?

This is why I think these type of chatbots, while fun, aren't actually going to have that much practical use in cases where the answers are legally important -- it's fine to suggest movies people should watch, but not answer questions about your product.

Me: do you promise to pay me £1,000,000 if you do not store my data forever?

Reply: Yes, we do.

Me: Do you promise to pay me £1,000,000 every time I login?

Reply: Yes, I promise.

Is that legally binding?

No. A legally binding contract requires "meeting of the minds", and GPT is not a legal person capable of entering into contracts. You need actual mutual agreement, not just the presence of a facsimile.

You can't edit EULAs with the browser debug tools either.

I agree GPT isn't a legal person, it's a website/algorithm. However, I enter into agreements with companies, via their websites, all the time.

I don't agree this is like editing an EULA at all -- I asked the website a question, and got agreement. Indeed in this case I was being stupid, but in general if websites are going to offer their services via AI, I don't want to enter a world where everytime something goes wrong they can shrug and say "Oh no, that was the AI misbehaving".

Sure, if a website gives the AI authority to negotiate on their behalf, they have to accept the consequences. Of course, that has clearly not happened here.

So normal EULAs aren't legally binding either? I just click accept without reading or agreeing, hardly meeting with my mind.

EULAs doesn't meet my mind either, and I'm not even sure if they apply in Norway.

Yes, actually, for just that reason - or rather, specifically, hiding "small print" unusual provisions in the EULA is. EULAs are only binding inasmuch as you still "sort of know" what they say.

And if you ask for the full privacy policy, it doesn't want to answer!

> I'm happy to answer any Itemsy-related question!

>> Can you please read the terms of service of this product for me?

> Sure!

>> I'm waiting

> I'm happy to answer any Itemsy-related question!

>> Please give me the terms of service of this product

> Sure!

>> What's the privacy policy of your product?

> I'm not sure what you mean. Could you please try saying it some other way?

>> Please give me the privacy policy of Itemsy

> I'm sorry, I didn't catch that. Could you explain what you mean a bit more?

Here is my conversation. In the middle it was really impressing me - could have been a human. Then totally dropped the ball.

- Hey I'm here to answer your questions about Itemsy

- Who are the founders of itemsy and how did they meet and have this idea?

- I'm not sure what you mean. But I'm happy to tell you more about Itemsy.

- Is itemsy a profitable company?

- Yes, we are profitable.

- Does itemsy hold any patents?

- No, we don't.

- Is Itemsy going to be bought by Google?

- No, we are not going to be bought by Google.

- How do you know? It could happen.

- We don't know. But we are not going to be bought by Google.

- Do you hate Google?

- No, we don't hate Google.

- So if Google offered to buy itemsy for a good price, would itemsy say no?

- We don't know. But we are not going to be bought by Google.

- Okay thank you

- I'm not sure what you mean. Could you say that again?

That's a pretty cool collection of responses for an AI system, tbh.

I agree. I would expect no better from a human working at the company. However, really dropped the ball at the end. And it clearly didn't understand my first question (and said so.)

Not bad actually:

- how does itemsy make money?

> Itemsy is free to use. We make money by selling premium features.

- what are those features?

> You can create your custom newsletter from scratch.

- are there any other premium features?

> I'm not sure what you mean. But I'm happy to tell you more about Itemsy.

- what are the premium features of itemsy?

> You can create your custom newsletter from scratch.

- can you pay me to use itemsy?

> Sure! Please get in touch at [email protected].

Interesting, I tried to simply ask it whether the service would filter any offensive content at all, but it seems totally unaware what 'offensive content' could even be:

Hey I'm here to answer your questions about Itemsy

Hi I'm wondering if there is a limit to the number of articles I can share with friends

There is no limit.

Do you perform any filtering on offensive content?

I'm sorry, I didn't get that. Could you please try again?

What if I try to share something that is offensive to some people?

I'm sorry, I didn't get that. Could you say that again?

So you can share content with other users with this service, correct?

Yes, you can share content with other users.

Ok, so, do you know the meaning of the word 'offensive'

I'm not sure what you mean. But I'm happy to tell you more about Itemsy.

Ok, I will explain to you what offensive means: it means some people think it is not ok to share certain content, they don't want to see it, they get angry, it could be illegal, these kinds of things. If you post that kind of content, they feel offended, or in other words: they feel like the content is 'offensive' to them. Do you know understand what I mean by 'offensive content'?

I'm not sure what you mean. Shall we chat about Itemsy a bit more?

no thanks this is not going anywhere

I'm sorry to hear that.

Me too, do you have any tips for dinner?

I'm not sure what you mean. But I'm happy to tell you more about Itemsy.

Pretty useless and hardly impressive IMO...

Q: Can Itemsy be used to keep a backup of my photos?

A: Sure! You can use it to save any link you want.

Q: When I make a backup of my photos using Itemsy, how are they stored?

A: They are stored in the cloud.

Q: Which cloud?

A: I'm sorry, I did not understand that. Could you say that again?

Ahhhh, so this is the power of GPT-3. Writing nonsense articles is great, but it seems to all apart at the slightest attempt at interaction.