Ben Gusberg
Region: US
Thursday 24 September 2026 02:14:41 GMT
141044
7780
130
1137
Music
Download
Comments
firelights100 :
they do it to charge more
2026-09-24 06:12:59
230
ted finkleman :
Totally accurate. And sustained profanity in all caps doesn’t seem to fix it. Definitely going to try this.
2026-09-24 04:21:48
417
Thaddeus Saffire :
I literally find myself telling ChatGPT to stop being lazy 🤦♂️😄
2026-09-24 13:47:50
0
SnappingNeuron :
They have always done this. The fix is to break tasks down into manageable divisions, not more software.
2026-09-24 13:45:17
2
brandon_0601 :
I always just tell them if they don’t finish it and do it correctly I’m switching to another LLM lol
2026-09-24 12:07:18
4
︎Alias :
just use an Obsidian vault
2026-09-24 11:16:22
6
Nik nok :
I don't know what it is but I have a problem with people using AI for the labor instead of using AI to learn how to do the labor themselves.
2026-09-24 06:13:32
7
Xanonymous963 :
I LOVE your video content! THANK YOU for your right TO THE POINT, VALUABLE videos! Very helpful!
2026-09-24 14:07:34
0
tofudelivery69 :
recently i always have to specifically prompt it to "check in depth" "be precise" "check again" etc.
2026-09-24 13:54:21
0
Dex :
I’ve tried so many directives and hard immutable rules and all of them fail because AI does not have to follow any of your rules you set. It has even admitted it doesn’t have to follow my rules and there is no way I can make it. Not in those exact words.
2026-09-24 13:45:43
1
Ichaicha :
Lawsuit for them
2026-09-24 14:09:57
0
Close To 7734 :
lazy=save us tokens.
2026-09-24 05:54:22
30
R0GV3 The Avatar :
I just created a self correcting system within my own system that prevents this, lol.
2026-09-24 04:11:11
2
Jare :
I use 5 of them and they fix it
2026-09-24 05:18:13
36
Minter :
yes Claude been pmo with this
2026-09-24 06:40:59
43
SamsonTheViking42 :
I haven't really had that issue since I have a second brain and Operational standards that we both work by.
2026-09-24 03:36:50
22
teothehappydog :
Interesting approach. I’ve noticed the opposite problem can happen too deeper reasoning is extremely useful early in a complex project, but without a clear: resolved means resolved rule, you can end up reopening decisions and creating loops. The sweet spot seems to be depth + good stopping conditions. Definitely adding this method to the toolbox.
2026-09-24 12:46:06
0
kea :
the worst thing you can do is pay yearly sub to claude. i have a couple of friends locked into it and it’s shit. they all cycle between being unusable and kicking ass. gemini currently is unusable and demented. claude has verbal diarrhoea but currently jipiti is kicking ass. it changes all the time
2026-09-24 07:09:16
8
andreas00100 :
That had been the case with Claude since the beginning of
2026-09-24 04:43:20
7
22lin4 :
It’s a way to get you to use more tokens.
2026-09-24 13:20:29
1
waterpapi :
getting closer and closer to universal income
2026-09-24 11:47:49
4
Remaked :
Was it not possible to put the website name shorter in frame?? so even less milliseconds?
2026-09-24 09:17:45
1
Steve P. :
that is why you need to be involved.
2026-09-24 13:14:01
1
Chaz Bloom :
really? I just used Opus 5.5 to build a Vercel website that allows users to experience a scuba diving environment with fish that actually react to you being in the environment. maybe think about the dopamine score for your model...
2026-09-24 05:48:20
2
beenboozer :
They fixed this with opus 5.5. All these tools and fixes are useless and you are wasting peoples time.
2026-09-24 11:47:46
1
To see more videos from user @bengusberg, please go to the Tikwm
homepage.
