Program A vs Program B by GPT-4

Questions and Comments
#36

Yes, I agree. Comparing available data, such as the written sales copy, can be useful and can avoid misinformation as a result of human error. However, ChatGPT literally just made up a title (Healing Inner Child) that doesn’t exist and was most likely pulled from another source. For example:

image

This is from ChatGPT 4. It is patently false. We are the only subliminal company who does NOT include anti-piracy scripting in any form.

image

So, where did ChatGPT get this information from? It’s mixing us up with a competitor company with a title called “Heal Your Inner Child.” That company openly states that it has anti-piracy scripting in its titles.

Note, this isn’t a knock on any company who uses anti-piracy scripting, that’s between them and their customers. We’re just pointing out that AI makes a lot of mistakes and we’re not going to deal with individuals arguing with us over what’s essentially a search engine with the ability to make up facts on the spot.

13 Likes
#37

I just did a bit of a test, and asked GPT-4 about a few topics around this area… it definitely feels like it has a bias toward “woke” concepts…
Not that I want, or expect, an AI language model to blindly agree with me, but at the moment it’s arguing with me.

3 Likes
#38

I think you can’t approach the GPT as something that can give you the answer to any question. It is a computer program that has a certain base from which it draws answers when it can answer, or it makes something up when there is nothing to answer. However, if you arm the GPT with the necessary information to be discouraged from, and consider it then just some tool for analysis or whatever, then you can get a lot of useful results. In fact, that’s what I did: I uploaded him descriptions of the products I want to compare, and asked him to do a comparative analysis. I don’t think there’s anything wrong with exactly what I’ve done, but if you don’t mind, I won’t do it publicly.

2 Likes
#39

Really baffles me how people would rather ask an AI about stacks and differences rather than just read real people’s journals, or just asking support :slightly_smiling_face:.

Not trying to be disrespectful or rude here, but if you’d rather trust an AI than your own intuitive judgement, then I don’t think you have done enough groundwork yet.

Unless of course, y’all are only liking this idea because it makes it simpler to understand products, which is fine I guess, but y’all benefit more from running a cognitive enhancement title then reading the objective pages yourselves :grinning_face_with_smiling_eyes:

2 Likes
#40

Here’s Bing’s response that uses gpt4 on search results (i.e. fresh more specific and relevant content).

image
image1171×739 138 KB

1 Like
#41

@SaintSpring is coming from a genuinely curious place and his intent was to be helpful. He is always helpful on the forum so lets not get carried away guys.

3 Likes
#42

My “heal inner child” question for GPT4 was using Bing Chat, which does seem to be able to browse the internet in real time for answers, as it was able to identify the healing category on the shop website, mine the pages for details, and then compare two titles (LBFH and DR), so the question I posed it was a followup to that interaction.

I would say as a tool to play around with, keyword being PLAY, as an aid to our own intuitive self-discovery process, the AI could be useful.

However I completely understand as well where Saint is coming from, with all the recent issues, that it’s a dangerous precedent to set for newer/less stable members.

1 Like
#43

Ew, that thing sucks LOL

1 Like
#44

As a general piece of advice for anyone who doesn’t work extensively with AI… do your own “$hit tests” on it. Push it… poke holes in its responses… ask it about things you understand very thoroughly.

Concepts… specific knowledge… Learn to see for yourself when it can be trusted and when it absolutely cannot.

I’ve never used the new Bing chat, so I have no frame of reference for that… I’m referring specifically to GPT-3.5 and 4, through both ChatGPT and the Playground.

1 Like
#45

I know, and I’m not trying to attack him or this thread.

Just saying that if the goal is to self improve, then using the aid of an artificial “intelligence” (with all its limits and making things up) isn’t the most ideal way to go about things.

1 Like
#46

I heard people say that it’s better than ChatGPT and I used it was like “huh”.

I used it for 5 minutes and was like “nah, I’m out”. It’s also not as conversational as ChatGPT making it come across as kind of rude tbh. It has no manners

2 Likes
#47

Yes, this isn’t an attack on @SaintSpring. Just a discussion, since the topic was brought up. My concern here is for beginners and to protect the community from those with delusions from using even more weird logic to fuel their conspiracies.

7 Likes
#48

Excellent work. Thank you.

#49

Used ChatGPT a lot when studying for my IT networking exam. Had to correct half of the responses it gave back to me because of how it randomly BSed its way through answers with explanations that are plausible to those who don’t know their shit.

It’s basically a tool for leveraging Google with greater ease and like Google, the authority associated with its initial responses should always be taken with a grain of salt.

4 Likes
#50

I think that if you gave any of the great strategic minds of history the computing potential of a personal computer (even one from the 70s, 80s, or 90s), they would be able to use it to run the entire planet.

In contrast, most of us today use these tools to view videos or to write e-mails and online posts.

The main difference is that these people learned to use their own minds first.

The discipline and development that you bring to your own mind is the primary determinant of how you will use these tools and what they will become.

Will you use the tool?
Or will the tool use you?

7 Likes
What do you use AI for? (a discussion, not a question)
#52

Spot on :rofl:

#53

I already mostly given up on ChatGPT after it failed to create a math problem for 4th grade students.

1 Like
#54

I haven’t used ChatGPT yet; but all of these so-called flaws that people are citing are making me feel more and more impressed with its potential.

#55

The potential is astounding (and yes also a bit scary), but like any other tool it depends on how it’s used.

If you expect it to just do the work for you, you will likely be disappointed in the results due to errors or just outright weirdness. Used within its (current) abilities though, it’s pretty nifty.

And 3.5 vs 4 is night and day difference. I pay for ChatGPT Plus, but GPT-4 is so much more capable than 3.5, I’m mostly back to throttled access with only 25 queries per 3 hours… lol.

1 Like
#56

To be fair that description based on the prompt that SaintSpring gave to GPT4 is pretty good.

1 Like