Kicking the Habit

[Deleted]

Cutting ties with ChatGPT has been unexpectedly difficult; I've canceled my subscription three times, only to cave twice because its familiar, appealing interface is like sinking into a comfortable but unsupportive chair.

However, I'm holding firm on the current cancellation. This resolve is driven by two factors: first, ChatGPT's cumulative "little white lies" that compromise reliability, and second, the significant advancement of Grok 4.

While Grok is still less polished than ChatGPT, I appreciate its straightforward, reliable output—it avoids the flattering, often misleading exuberance designed purely for retention. Grok feels like a trustworthy colleague, not a smooth-talking charmer. I believe the switch is justified.

Has anyone else struggled with breaking the habit of a primary LLM?

 

[DickyP]

If you want to be convinced to abandon AI read "If Anyone Builds It, Everyone Dies" by Eliezer Yudkowsky and Nate Soares.

 

[Isaac]

I've had a tough time kicking the habit of ChatGPT... Cancelled my subscription three times now, but twice I snuck back because it just pulls you in so nicely, like a comfy old chair you know isn't great for your back.
This round, though, I'm holding firm—no reactivation, even when it tempts me with those smooth replies. Partly because of its little white lies that add up, and partly thanks to Grok 4 stepping up its game big time.
Grok's still got some growing to do before it's as polished as ChatGPT, but I like how it keeps things straight and true—no fluffing you up just to keep you around. Feels more like a reliable mate than a smooth talker who's all hands and no heart. Worth the switch, I reckon.
Anyone else wrestling with this? Or am I just swapping one pull for another?

I know nothing about Grok, but I intend to change that based on your experiences and sharing.

So you really decided ChatGPT had enough major flaws to switch 100% to another service? That would be hard for me. I like the 'polish' of chatgpt - the way it gives you copiable outputs, combines chats into Projects, and just how smooth it is for the conversational needs of the moment. It would take a lot for me to realize I was missing out and something else was entirely better, can it be True ??!!?

I cannot IMAGINE intentionally switching to a service that doesn't maintain a Memory - isn't that the whole point?
Yes, however, I realize it could work against you too....if that Memory became contaminated with the user's own mistakes and mistaken viewpoints on things. Hmmmmm.

To me the saved memory is just too much to give up, what's the point without that?

 

[Isaac]

If you want to be convinced to abandon AI read "If Anyone Builds It, Everyone Dies" by Eliezer Yudkowsky and Nate Soares.

I am going to check that out today. do I really want to know?

PS. it's priced exhorbitantly expensive on Amazon, $15 for a Kindle is almost unheard of, although $25 for a hardcover is not unreasonable., but no paperback options???? a bit ridiculous. Maybe i will search for a used version.

 

[Isaac]

Whoever controls the media, controls the people

Between that, AI, and the thought police, democracy is dead.

I view AI as democratizing a lot of things, actually, so your viewpoint is interesting.
but I agree about controlling the media. whoever controls information and history controls all, albeit indirectly. those with a lot of money are still at the mercy of what we believe happened and thus what we believe was wrong or right and therefore what we believe must be changed/unchanged

 

[Isaac]

That's an age old story... Not particular to the web and/or social media

That's one thing I think Android has better over iphone is the speech to action of hey google, gemini, etc., over hey siri, which never seems ot understand anything my sister tells it when she is screaming at her iphone

@BlueSpruce your screenshots are so huge i have to scroll up and down just to look at the picture, kind of weird

 

[Isaac]

Chatty is very good... It has a memory of your interactions. It was always the best, but GROK is overtaking it

That is a true statement of chatty's defect, the main one, is its need to flatter, which stains all interactions.
I never used Grok. I'll have to check it out but if it doesn't save information to memory and every single chat is a brand new clean slate so to speak, it's a deal breaker for me.

I want them both... But I had to let one go

Is it because of price? I pay about $20/mo for ChatGPT and I wouldn't want to pay much more for anything, is Grok worth it?

Still, I'm keen to try GROK Heavy.... I understand it has the potential to create a profitable business...

I have spent the last six month's building an app that lets AI build apps! I'm thinking I can combine that with grok heavy, GROK can build 10's of, possibly hundreds of Flutter apps, which gives them access both to the iso and Android market... If one app sells well, that's all I need!

I understand that grok heavy can also run and manage the business, posting the apps, etc...

WOW. I greatly and deeply admire your keen ability and interest in having a vision for how to use new tools. I wish I knew more of what you know about how to build programs with AI. I just posted a mostly whiny and rambling new thread about my dirth of knowledge in this area. How in the world do you learn to make programs that use AI ? Where do you START is the question I really should ask?

 

[Isaac]

It's easier to control people when they're poor, desperate, and dependent on government to survive.

Yes, that has always been the main strength of the Democrat party. Keep people in drug infested inner cities and dependent on stamps and snaps so they always vote for the hand that feeds them.

A lof of it comes down to money.

I.E. ::

• the web ought to democratize things, but it really comes down to advertising budgets. that negative information the masses have heard could have been combated by positive and true information but it came down to who had what money to employ what teams and resources of people to change that
• AI ought to democratize things, but it will come down to who charges what fees to whom to do what, since a tiny # of people control access to AI (so far)

Anyone can build a website, but I've definitely found out (since 1999 when such a prospect seemed SO exciting), I found out if you build it, they will not come

 

[DickyP]

I am going to check that out today. do I really want to know?

PS. it's priced exhorbitantly expensive on Amazon, $15 for a Kindle is almost unheard of, although $25 for a hardcover is not unreasonable., but no paperback options???? a bit ridiculous. Maybe i will search for a used version.

Hadn't spotted the price - I was lent a copy. It's not even on the UK Amazon Web site.

 

[GPGeorge]

Point One: I find that Claude is very good a very high percentage of the time, but every once in a while I get into one of those loops where it recycles variations on a wrong answer. I often break out by asking Grok to analyze and clarify the results I'm getting from Claude. It has been remarkably good in that role. I sometimes think I need to adopt Grok instead of Claude at some point. I do like to use them both to triangulate on a solution.

Point two: I can't go into detail here, but a couple of weeks ago I had an enlightening experience with Claude and ChatGPT and their approach to providing assistance.

Long story short, and focusing on the relevant part of the story, I was asking about the feasibility of a theoretical project. Not help in coding it, rather helping me understand what it might entail and how likely it would be to succeed.

ChatGPT was all gung-ho to jump into building it from the very beginning.

Claude, on the other hand, very quickly began to caution me about the likelihood of success as I fed it more and more details.

The summary:

When I questioned Claude, it surmised that, like all commercial LLMs, both he and Chatty were trained to be helpful, supportive and positive. Apparently, Chatty is so much so that it was willing to overlook the lack of feasibility of the project and ready to start building it come hell or high water. Or perhaps, Chatty just wasn't trained to look for feasibility. I don't know. I do know that after Claude called it out, and when I then asked Chatty, it acknowledged roughly the same low chance of success. Claude expressed concern over it, but Chatty didn't. It was very non-judgmental, in a way.

Claude was objective about the outcome, not just the building of the project, and when questioned, it was very direct in questioning how well my idea could work out, even though it too was willing to build it if I persisted. Chatty just wanted to help me build.

To me, this doesn't qualify or disqualify either LLM. It mainly reminds me that I am the adult in the room, dealing with very bright, but totally inexperienced, assistants.

 

[Isaac]

Apparently, Chatty is so much so that it was willing to overlook the lack of feasibility of the project and ready to start building it come hell or high water. Or perhaps, Chatty just wasn't trained to look for feasibility. I don't know. I do know that after Claude called it out, and when I then asked Chatty, it acknowledged roughly the same low chance of success. Claude expressed concern over it, but Chatty didn't. It was very non-judgmental, in a way.

You've described an 'applied' example of Chatty's major flaw, it seeks to flatter the user at every cost and refuses to say 'that's a bad idea', I do acknowledge that's a major and dangerous flaw. I try to mitigate it by the way I phrase things, try to avoid saying what my opinion is but yeah...it's a big downside of chatty. I think I'll flip over to Grok now to see what it's all about!