The most prestigious law school admissions discussion board in the world.
|
 |
 
|
"AI" is a complete scam
Poast new message in this thread
Date: August 13th, 2026 5:34 PM Author: white nibblets (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50066182) |
Date: August 13th, 2026 5:37 PM Author: Kink-friendly indigo sanctuary circlehead i think for consumer shit like i dont see CC or banks using it really, its just BS to make it harder for u to contact actual indians in bangalore for CS
but structurally comps will use it to screw over goys, improve productivity in back office shit etc. so that shld improve corporate profits a lot |
 |
Date: August 13th, 2026 5:49 PM Author: obsidian bawdyhouse It doesn’t do shit other than write emails for people who can’t write or get cut rate IT shit work that you spend months explaining how to get it working, which you would otherwise have to do the same with third world pajeets (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50066209) |
|
Date: August 13th, 2026 6:02 PM Author: Vigorous ungodly ticket booth kitty cat It can do legal research and drafting now (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50066219) |
|
Date: August 13th, 2026 9:23 PM Author: insecure mind-boggling shitlib market unless u people are like landscapers i have no idea what u think AI cant do. it does everything u can do on a computer but better most of the time unless youre actually senior and actually smart (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50066701) |
|
Date: August 13th, 2026 11:16 PM Author: white nibblets An associate used her Claude Gay.Pro to synthesize an Excel worksheet with a bunch of .pdfs, to match up the .pdfs by Bates number to the Excel entries.
It produced a great beautiful spreadsheet that had tons of additional rows of data extracted from the .pdfs. But then when I started looking, all the Bates cites were fucked up; like consistently off by either 2 or 3 numbers in one direction or the other. And around 15% of the other entries were wrong. So basically useless because you'd have to check 100% of the entries, or not use it. |
|
Date: August 14th, 2026 7:02 AM Author: Haunting maize idea he suggested pozpig But does anyone see this and think “it won’t get better?” Are the critics assuming the fuzziness will forever be there? Because it seems reasonable that those sort of problems are going to be ironed out pretty quickly. |
|
Date: August 14th, 2026 7:21 AM Author: Cracking Opaque Potus Using AI for law is different than every other industry
There is symmetrical cost/risk for lawyers using it vs asymmetrical upside for everyone else since we can't afford to be wrong one time
The name of the game is verification gating and cross-model validation which addresses your last point. Have another model independently check it not yourself |
|
Date: August 14th, 2026 9:34 AM Author: obsidian bawdyhouse yep, lawyers are the only ones who need to produce documents and numbers that aren't bullshit. everyone else is just making pretty looking charts.
you fags write those 50+ page "briefs" specifically because you know no one will ever read them so if you say something stupid or flat out wrong it will get lost in the weeds |
|
Date: August 14th, 2026 7:29 AM Author: Galvanic thirsty black woman Why is it that "professionals" seem to exclusively use AI to read pdfs (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067248) |
|
Date: August 14th, 2026 9:41 AM Author: Mahogany base because we spend an inordinate amount of time doing exactly this, and it *seems* like AI would be able to do this with ~100% accuracy, within seconds. but it cant. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067412) |
|
Date: August 14th, 2026 7:40 AM Author: Chestnut Bat-shit-crazy Gas Station Generalized Bond Oh if only we dumped a few more trillion into data centers and burned off enough natural gas to power the country for twenty years, only 10% of your entries would be dogshit. Nothing wrong with AI. Just needs more scale! (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067264) |
|
Date: August 14th, 2026 4:34 AM Author: obsidian bawdyhouse I run a division heavily responsible for data analysis. We’ve been doing an AI pilot for the last few months with multiple use cases, regular meetings, etc and did a final debrief yesterday afternoon. The consensus across the board was that this shit is retarded and gay and useless (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067170) |
|
Date: August 14th, 2026 4:45 AM Author: Bronze field explain how it did data analysis wrong (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067172) |
|
Date: August 14th, 2026 7:01 AM Author: Haunting maize idea he suggested pozpig (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067221) |
|
Date: August 14th, 2026 9:35 AM Author: obsidian bawdyhouse it gives shitty information that takes longer to unfuck than just doing it yourself (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067401) |
|
Date: August 14th, 2026 7:31 AM Author: Galvanic thirsty black woman Sounds like youre either lying or some wordcel retard working with retards. "business intelligence" and the like is going to disappear even faster than programming (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067250) |
|
Date: August 14th, 2026 9:24 AM Author: obsidian bawdyhouse One person's notes:
Leveraged Copilot to quickly aggregate and synthesize market data to accelerate analysis (1) Retail pricing project: used Researcher Agent to gather pricing data from competitors, enabling faster turnaround on pricing insights and shifting time spent on data gathering to decision quality analytics (2) Market share analysis: used Researcher Agent to compile external insights on industry and market benchmarks to improve speed and consistency of data used in market share analysis of key training products. Resulting in faster synthesis of industry benchmarks and allowing greater focus on strategic implications
Challenges: -> While AI drastically improves speed of data collection, hallucinations still require thorough vetting of reference data and sources to eliminate bias from the results. -> Using AI for analysis has not proven more effective or efficient for routine analysis. For new projects, determining effective prompting and checking outputs can take longer than building the analysis without utilizing AI\
TLDR: AI is a load of shit. The meeting I mentioned was over in under 10 minutes and everyone had the same consensus. |
|
Date: August 14th, 2026 9:45 AM Author: Galvanic thirsty black woman I already addressed this with "wordcel retard working with retards". Thanl (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067420) |
|
Date: August 14th, 2026 10:23 AM Author: obsidian bawdyhouse wtf is a "wordcel"? normal people don't talk like that (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067540) |
|
Date: August 14th, 2026 10:43 AM Author: Galvanic thirsty black woman (literal pumo posting on a 23 year old web 0.7 bbschat pretending he "doesn't know internetspeak because that's fag shit for fags") (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067557) |
|
Date: August 14th, 2026 11:24 AM Author: obsidian bawdyhouse I genuinely have no idea what a "wordcel" is. If it means you use ms word as a core function of your job then no that's not me. also not one of the those retarded, usually indian, codemonkeys I sometimes have to boss around when we want shit to get done. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067633) |
|
Date: August 14th, 2026 12:25 PM Author: Galvanic thirsty black woman So what is it exactly that you "do"? (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067756) |
|
Date: August 14th, 2026 12:27 PM Author: obsidian bawdyhouse why don't you tell me wtf a "wordcel" is first? (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067760) |
|
Date: August 14th, 2026 1:34 PM Author: Galvanic thirsty black woman A Liberal Artist. Now, what is it that you "do"? (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067893) |
|
Date: August 14th, 2026 1:51 PM Author: obsidian bawdyhouse i don't do liberal arts. also ppl are not dead because of me. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067926) |
|
Date: August 14th, 2026 10:45 AM Author: awkward stead trump supporter he's a chopped unc, speaking an outdated language to himself (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067563) |
|
Date: August 14th, 2026 11:37 AM Author: insecure mind-boggling shitlib market sol 5.6 can be a pain in the ass and do things like re-run th same control set 10000x, but i dont know if ive ever seen it hallucinate like it was gpt3
i think ur guys are full of shit and bad prompters and low iq |
|
Date: August 14th, 2026 11:40 AM Author: Iridescent Floppy Immigrant codex sol 5.6 hallucinates very little and even when it does it is trivial to mitigate. You can even spawn subagents that use tools to double check everything and make sure it is exactly right. like if for the old issue of ai "hallucinating sources" you can put all of the real sources in the repo and it will use an agent and a tool to inspect them and make sure they are exact (can do this with "data" too) and it isn't particularly hard to do. he's either just making shit up or low iq. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067665) |
|
Date: August 14th, 2026 11:42 AM Author: obsidian bawdyhouse AI fags are worse than climate change fags in that they can't even explain wtf the point is for any of this shit. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067674) |
|
Date: August 14th, 2026 11:43 AM Author: insecure mind-boggling shitlib market ...to do all computer work
remember email girls?
jesus fucking christ u people are so afraid of reality |
|
Date: August 14th, 2026 11:50 AM Author: obsidian bawdyhouse what computer work do you think people with actual jobs are doing? your understanding of reality seems more like a collection of tiktoks and internet memes (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067690) |
|
Date: August 13th, 2026 5:38 PM Author: charismatic alcoholic property you're just mad retards are using it to PWN you in court now (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50066189) |
|
Date: August 13th, 2026 5:44 PM Author: deep snowy parlor If you think AI is a scam, wait til you see what we have next (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50066202) |
|
Date: August 13th, 2026 5:44 PM Author: Stimulating library jewess you really are not an AI guy are you epah (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50066204) |
|
Date: August 13th, 2026 6:01 PM Author: white nibblets I just like to get attention on xo. No one even responds to TDS spam anymore. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50066218) |
|
Date: August 13th, 2026 6:02 PM Author: Vigorous ungodly ticket booth kitty cat post cock pics then. erect ones. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50066220) |
|
Date: August 13th, 2026 9:08 PM Author: disturbing degenerate (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50066670) |
|
Date: August 14th, 2026 5:29 AM Author: Heady Place Of Business Menage THIS. FURKING THIS. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067181) |
|
Date: August 13th, 2026 8:59 PM Author: contagious scourge upon the earth corner (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50066652) |
|
Date: August 13th, 2026 9:21 PM Author: charismatic alcoholic property your event horizon-esque child multilations are forevergreen to generate responses (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50066695) |
|
Date: August 13th, 2026 9:31 PM Author: Cracking Opaque Potus Neither "AI" nor "scam" has any commonly accepted meaning in the year 2026 that would enable anyone intelligent to respond to you (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50066721) |
|
Date: August 14th, 2026 5:29 AM Author: Heady Place Of Business Menage Iverson was overrated but I wouldn't say he was a complete scam. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067180) |
|
Date: August 14th, 2026 7:03 AM Author: Haunting maize idea he suggested pozpig AI never got a team or organization behind him. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067225) |
|
Date: August 14th, 2026 6:37 AM Author: razzmatazz pearly school cafeteria milk its fine who cares |
|
Date: August 14th, 2026 6:45 AM Author: awkward stead trump supporter yet the pillars of America's economy and financial resources hinge on AI-- when the return on AI investment is negative (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067201) |
|
Date: August 14th, 2026 7:32 AM Author: Galvanic thirsty black woman Aren't you like 55 by now (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067251) |
|
Date: August 14th, 2026 7:44 AM Author: awkward stead trump supporter no way Jose, I'm no chopped unc like u (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067282) |
|
Date: August 14th, 2026 7:44 AM Author: Chestnut Bat-shit-crazy Gas Station Generalized Bond My favorite part is all those dorky programmers telling me a year ago that AI makes them 1000x more productive. “Just you wait to see the apps that will come from this!” Well the top ten app list looks like the same shit that’s always been there. And I don’t see a lick of difference in anything I use. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067280) |
|
Date: August 14th, 2026 7:46 AM Author: Haunting maize idea he suggested pozpig They just mean that they can make the same shitty apps 100x faster. A lot of coding is unpleasant grunt work. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067283) |
|
Date: August 14th, 2026 7:48 AM Author: Galvanic thirsty black woman He made his anecdote up (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067287) |
|
Date: August 14th, 2026 11:44 AM Author: Iridescent Floppy Immigrant (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067678) |
|
Date: August 14th, 2026 7:55 AM Author: Chestnut Bat-shit-crazy Gas Station Generalized Bond So it’s the equivalent of excel for programmers. Wow big whoop! It’s clearly not as game-changing as these Silicon Valley vampires would have you believe. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067293) |
|
Date: August 14th, 2026 7:58 AM Author: Galvanic thirsty black woman Lawyers are all gonna die (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067296) |
|
Date: August 14th, 2026 9:39 AM Author: obsidian bawdyhouse it's the equivalent of macros and powerBi for excel users (i.e. - something that looks fancy but no one effectively uses and something being forcememed as an improvement even though it does the same thing with the same amount of effort while losing connectivity to the underlying data) (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067409) |
|
Date: August 14th, 2026 10:35 AM Author: Crawly Address it really is - the issue is all the top apps / players are stuffing ai into their platforms and almost all of it is shit "Meta AI" Google "Smart Reply" etc and this is what the majority of people are experiencing.
|
|
Date: August 14th, 2026 11:46 AM Author: insecure mind-boggling shitlib market yeah man guys at my high school spun up $1b enterprises in one year all the time nbd
(opus 4.5 came out November 24, 2025) |
|
Date: August 14th, 2026 11:59 AM Author: obsidian bawdyhouse no one knows what any of this shit is you fag and no real company is going to adopt any of it. it's literally just for chinks and turds who do nothing but write code and can't write coherent engrish (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067703) |
|
Date: August 14th, 2026 12:28 PM Author: Iridescent Floppy Immigrant You don't think it's a big deal that a machine is far more intelligent than you? (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067761) |
|
Date: August 14th, 2026 12:29 PM Author: obsidian bawdyhouse if it's so smart then why can't it do even basic things? (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067765) |
|
Date: August 14th, 2026 12:28 PM Author: Glittery Alcoholic Range Preventive Strike it's not really obvious why it would dramatically change the user experience or the amount of software available. the existing software satisfied most needs. the point is you don't need as many people to work in SWE or data analysis or other fields where agentic models can easily do many things. anthropic currently has an annualized run rate of around $75 billion, OpenAI at $40 billion plus all the other model providers. obviously enterprises are finding it increasingly useful and are willing to pay for it. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067762) |
|
Date: August 14th, 2026 9:44 AM Author: Stirring flickering brethren institution I need it to not collapse so I can retire. Please keep pumping conjured money into this ecosystem so numbers continue going up (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50067418) |
|
Date: August 14th, 2026 4:11 PM Author: white nibblets No way am I reading all that ^^ (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50068332) |
|
Date: August 14th, 2026 5:36 PM Author: sooty vibrant mexican Crazy how satisfying it is when I email someone and get a responsive, thoughtful reply. Those types of responses are so sharp, insightful, to the point etc it really underscores how abysmal and BS this meandering, nonsubstantive AI slop stuff is. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50068505) |
|
Date: August 14th, 2026 5:38 PM Author: white nibblets I think that reaction makes sense, but it may say as much about what AI is optimized for as it does about the quality of AI itself. A thoughtful human email usually comes from someone who knows you, understands the context, has a specific reason for responding, and has spent some time deciding what actually matters. AI, by contrast, is often optimized to be helpful across an enormous range of questions, which can produce a lot of qualification, hedging, repetition, and generic connective tissue. It can sound fluent without necessarily being substantive. There’s also something psychologically different about knowing that a person actually read what you wrote, formed an opinion, and chose their words in response to you. The irony is that AI can produce genuinely sharp, insightful, concise writing when pushed toward that goal—but its default tendency is often to say everything that might be useful rather than identify the one or two things that actually are. (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50068508) |
|
Date: August 14th, 2026 5:48 PM Author: Galvanic thirsty black woman how drunk r u rn talking to mr. gippity (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50068517) |
|
Date: August 14th, 2026 5:49 PM Author: white nibblets ? (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50068519) |
|
Date: August 14th, 2026 5:50 PM Author: Galvanic thirsty black woman ¿ (http://www.autoadmit.com/thread.php?thread_id=5892833&forum_id=2\u0026show=6hr#50068525) |