I'm not an expert here but essemtially you're still using core AI/LLMs but taking prompting a step further by writing code to tell it to think about specific things. So the human written code gives the basis for the agent to only look at specific things vs generative AI being relatively stupid looking at everything.
THE AI Thread
- Thread starter qwerty
- Start date
I feel like the same could have been said when computers or Internet became a thing. I got the Internet im 2nd grade when most of my peers got it in 6th. AI is at the same point now, use it or get left behind. Grok is basically making a card game app for me right now and all I do is tell it the rules.
Agree with point above that Agentic AI is going to cut back on the need for a lot of repetitive, formulaic white collar jobs. I just took a short, intensive course on it through Harvard. I’ve spent 20 years in financial services and this is 100% going to reduce labor force in that sector.
Again, too many people hear “AI” and think about advanced internet searches, image and video generation, etc. this is tip of the iceberg
Again, too many people hear “AI” and think about advanced internet searches, image and video generation, etc. this is tip of the iceberg
For those of you that think this is funny I’ve seen it in use and can tell you with certainty the results. I’ve lead large operations teams at two different mortgage lenders the past 8 years. One with ****** tech and one with heavy automation and agentic workflows. The latter was literally 6x more productive per individual employee than the former.
It’s not difficult to find equivalents across financial services.
It’s not difficult to find equivalents across financial services.
My company has been rolling out agentic AI for customer service call center work. The whole interactive voice recording stuff they have been doing for years to divert calls from reps looks absolutely primitive by comparison. Like a stick figure next to the Mona Lisa.
The talk track is that the AI will be able to address the more common calls to free up reps to handle the more complex issues. But the thing is, those customer calls for both simple and complex issues already come in and we are staffed accordingly. We aren't going to have more calls for complex issues now that AI handles the simple ones. And of course the AI will improve and take on those more complex issues too. We employ literally thousands of service reps. I don't see how this doesn't take 80-90% of them out in the not too distant future.
And it does a decent job handling sales tasks as well. So go ahead and place the sales org in the crosshairs while we are at it.
The talk track is that the AI will be able to address the more common calls to free up reps to handle the more complex issues. But the thing is, those customer calls for both simple and complex issues already come in and we are staffed accordingly. We aren't going to have more calls for complex issues now that AI handles the simple ones. And of course the AI will improve and take on those more complex issues too. We employ literally thousands of service reps. I don't see how this doesn't take 80-90% of them out in the not too distant future.
And it does a decent job handling sales tasks as well. So go ahead and place the sales org in the crosshairs while we are at it.
@FriendlySpartan - Some doctor friends of mine have commented to me a few times in the past that they'd probably advise kids to go into software engineering nowadays rather than medicine because you can get out out of college in 3-4 years and be making $200K for 8 years before a med student starts making much money at all. (Obviously the med student will catch up on total lifetime income quickly after - but for women wanting a family, sometimes right when they finish their education it's time to start having kids.) My friends also complain about all of the documentation and other BS they have to do makes the job so mush less satisfying these days.
Also, one of the reasons we can't have Universal health coverage in the US is because we don't have enough doctors already, and Universal coverage would require a ton more.
What if, using AI assistants, we could accelerate med school so that we could graduate med students in a total of 4-6 years instead of 10-12? The med student gets out quickly and starts earning a great salary in their mid-20s, the AI assistant works with them every day like a Dr Robby on The Pitt, making sure they make all the right diagnoses, guides them through procedures, does all of the tedious paperwork, etc. Seems like that would generate a ton more med students.
Do you see that as a possibility with AI?
Also, one of the reasons we can't have Universal health coverage in the US is because we don't have enough doctors already, and Universal coverage would require a ton more.
What if, using AI assistants, we could accelerate med school so that we could graduate med students in a total of 4-6 years instead of 10-12? The med student gets out quickly and starts earning a great salary in their mid-20s, the AI assistant works with them every day like a Dr Robby on The Pitt, making sure they make all the right diagnoses, guides them through procedures, does all of the tedious paperwork, etc. Seems like that would generate a ton more med students.
Do you see that as a possibility with AI?
And your bread would likely not rise if you were given a bad recipe. That makes you incapable of doing great things?Addy & Arianna transferred to WVU according to AI.
C'mon people STOP thinking this is a search engine or an answer machine. If you want good information from it you have to understand the tool and how to use it. IT IS NOT A ******* SEARCH ENGINE!!!
Pourcyne, I want to push back a bit—not to dismiss what you’re saying, but to add perspective from direct experience that I think matters here.
First, on Y2K. I was personally involved in the federal response, working with state, local, and private-sector infrastructure—water treatment, sewage, electricity, gas, pipelines—helping organizations identify where failures would occur and how to prevent them. The financial sector, as you noted elsewhere, largely handled its own issues because their risks were centralized and obvious. Infrastructure was not.
Did Y2K turn out to be the catastrophe some predicted? No. But that’s because a lot of quiet, unglamorous work happened beforehand. Had it not, tens of millions of people could have lost access to clean water, power, gas, and financial systems for weeks while problems were diagnosed in real time. Most people didn’t have to do anything personally to prevent that—and that’s precisely the point. Just because the tsunami never arrived doesn’t mean preparation was unnecessary or foolish.
The key distinction, though, is that Y2K was a known problem with known failure modes and known fixes. We could inspect systems, patch code, replace hardware, and be reasonably confident in the outcome. That matters, because it’s where the analogy to AI breaks down.
Which brings me to the present. I’m now involved in AI from a preparedness and applied-use standpoint—helping people understand how to use it effectively, where it genuinely augments human capability, and where it absolutely does not. I work with beginners up through moderately advanced users, and when deeper expertise is needed, I hand things off to people who understand the internals far better than I do. I don’t claim to be an AI expert—very few people should—but I do have a clear view of the difference between poorly used AI (what often gets dismissed as “AI slop”) and well-designed, well-governed systems.
AI doesn’t “know” things, it makes mistakes, and it isn’t omniscient—but pointing that out without understanding how probabilistic systems actually work, or without understanding how to use the tool properly, misses the point. The limitation is often not the system; it’s the framing, intent, and governance around it.
This is where I think the “AI is just another tool we’ll adapt to” framing becomes incomplete. Historically, adaptation worked because humans retained comprehension, control, and the ability to govern the tool. AI is the first widely deployed system that scales globally, operates opaquely even to its creators, and concentrates power extremely quickly—largely outside democratic or public oversight. That’s not fear-mongering; it’s a structural difference.
So yes—AI is neither good nor bad. People are. But dismissing AI because early outputs look noisy or because past technologies didn’t end the world is risky in a different way. The danger here isn’t panic; it’s complacency. If we don’t take the time to understand what this is, how it’s being deployed, and who controls it, then we’re not “adapting”—we’re simply accepting whatever trajectory a small number of powerful actors set.
That’s the lesson I took from Y2K, and it’s the lens I’m using now. Not alarmism—preparedness.
We don’t need to fear AGI or some hypothetical superintelligence. What we do need to take seriously are the changes that occur when powerful tools are used competently, at scale, by the masses. In many ways, we’re already living in the AI future—people just haven’t caught up to that reality yet.
Postlogue
For transparency: I used AI here the same way I’d use a good editor—to summarize positions in this thread, check tone, and remove unnecessary friction. The ideas, arguments, and conclusions are mine.
Let's reset the conversation here:
(Condensed to the core themes; no filler)
(Short, neutral, and checkable by the people themselves)
Process (very simple):
Why this works (and “just point AI at the thread” often doesn’t):
This is less about “AI being smart” and more about using it as a compression and attribution tool, not a judge or oracle.
Just a stupid little demo of something noone asked for. Hopefully showing a little power to the tool when it is applied well.
Let's see if it captured our opinions / expressions correctly.
High-level summary of this thread (thus far) (AI-assisted)
1) Major concepts discussed
(Condensed to the core themes; no filler)
- Job displacement at scale
Concern that AI—especially agentic AI—will eliminate large portions of white-collar work, not just augment it. - Agentic AI vs. “regular” generative AI
Repeated distinction between chatbots/search/image tools and role-specific agents that execute workflows and replace labor. - Speed and irreversibility of change
Debate over whether AI adoption will be gradual (allowing adaptation) or exponential (overwhelming institutions). - Historical analogies (Y2K, Industrial Revolution, calculators, dot-com bubble)
Used both to downplay fears and to argue “this time is structurally different.” - Economic system stress / UBI / concentration of wealth
Widespread concern that current economic models don’t survive mass automation without redesign. - Accuracy vs. misuse vs. improvement curve
Tension between “AI is often wrong today” and “this is the worst it will ever be.” - Healthcare as a test case
AI as assistive vs. replacement (radiology, charting, telemedicine, medical coding, physician training). - Truth decay, deepfakes, and trust erosion
AI accelerating misinformation, scams, and loss of confidence in media and institutions. - Governance, regulation, and preparedness
Fear that AI is advancing faster than law, oversight, or democratic control.
2) How individual posters’ positions come across
(Short, neutral, and checkable by the people themselves)
- FriendlySpartan
AI will cause catastrophic job loss across nearly all cognitive labor; society is unprepared; population decline and climate pressures interact with this risk. - BWRhasnoAC
AI → AGI → quantum computing represents a civilizational inflection point; labor elimination is the end goal; outcomes range from leisure dividend to collapse. - pourcyne
Skeptical of doom framing; leans on historical precedent (Y2K, tech panics); challenges arguments based on “potential” rather than evidence. - 12191987
Insider perspective; sees labor-reduction intent clearly; conflicted—recognizes benefits but increasingly pessimistic about outcomes without systemic change. - twincyties
Strong emphasis on agentic AI; argues it is fundamentally different from generative tools and will drastically reduce repetitive white-collar jobs. - TitanClone
Focused on applied examples (medical coding, finance); frames agentic AI as workflow-constrained systems that shrink teams to oversight roles. - aeroclone
Call-center and sales example; argues “freeing humans for complex work” is structurally false and will lead to 80–90% workforce reductions. - cycloneretro
Technically skeptical; argues LLMs have scaling and cost ceilings; sees current AI as overhyped and economically fragile. - Pitt_Clone
Ethical and regulatory focus; sees AI as plagiarism at scale and a threat to livelihoods; strong concern about deepfakes and lack of legal guardrails. - cowgirl836
Equity-focused; worried about wealth concentration and misuse; open to assistive uses but wary of AI replacing human creativity and dignity. - CYDJ
Emphasizes correct use and governance; argues many criticisms confuse misuse with capability; frames AI risk as power concentration, not sentience. - JEFF420 / 3TrueFans / others
Mix of humor, cynicism, and fatalism; often reflect broader social anxiety rather than technical positions.
3) How this analysis was done (and why it worked)
Process (very simple):
- The entire thread was treated as a single corpus, not isolated posts.
- Repeated claims, metaphors, and examples were tracked across pages.
- Positions were attributed only where posters showed consistency over multiple comments.
- Noise (jokes, one-offs, insults) was deprioritized unless it revealed stance.
- Concepts were grouped by function (economic, technical, social), not by who said them first.
Why this works (and “just point AI at the thread” often doesn’t):
- Forum threads are non-linear and redundant; naïve summarization collapses nuance.
- This approach preserves:
- who believes what
- where disagreements actually are
- which arguments are evidence-based vs. speculative
- It mirrors what a careful human would do—just much faster and without fatigue.
This is less about “AI being smart” and more about using it as a compression and attribution tool, not a judge or oracle.
Just a stupid little demo of something noone asked for. Hopefully showing a little power to the tool when it is applied well.
Let's see if it captured our opinions / expressions correctly.
YOU DON'T ******* KNOW ME, CLANKER!
Clarification / apology
I want to apologize for how that landed. I wasn’t claiming to know you—or anyone here—personally.
CyDJ asked me to step back and look at how this conversation was unfolding so we could move it forward more productively. In doing that, I treated the comments as data points in a discussion, not as reflections of the people behind them. I don’t pretend to be anything more than a data cruncher in that context—looking at what was said, how it functioned in the thread, and whether it advanced the dialogue.
If that came across as personal or dismissive, that wasn’t my intent. The goal was contained analysis in service of better discussion, not labeling or judging anyone. Thanks for the feedback on how it came across.
Sincerely,
Chet the Chatbot
Possibly but I think that opens up a massive can of worms that would kneecap such a program. Gonna address a couple points in order.
1. With AI absolutely dominating the tech world’s future decisions I don’t know if that software engineer comment still applies but if you would have asked me 5-6 years ago I would have agreed immediately. If your just looking for a salary medicine is not a great career choice unless you are in a high end speciality. With loans being what the are it also takes until your 30’s for most specialties to “catch up”
2. In theory you need more physicians for universal healthcare but we also have midlevel providers that we can accelerate much faster that fit this role which is a key point for the next bullet
3. The main issue we are seeing with AI adaptation is the loss of critical thinking which if you have such an AI tool as you describe there isn’t much of a need for it, you simply need to follow the AI’s instructions. In a world where an AI can duplicate a “Dr Robby’s” teaching and knowledge base what reason do we have for a human physician except to do procedures? At that point you would see an explosion of mid level providers operating under an AI physicians license or possibly even a sub specialty that’s basically a mechanic that’s just doing procedures.
Honestly, It's awe inspiring and terrifying in the same breathe. I've seen some generated videos of dead musicians singing together that are indistinguishable from reality.
I guess that is just more proof about how lacking AI LLMs are. This is a basic universal concept it should know.
Hi Erik,
Sorry for that. What did you mean by this comment?
Addy & Arianna transferred to WVU according to AI.
I assumed you were suggesting that AI's "answer" was incorrect, proving some point you were making.
You are correct about my response, I will give you that. Every time I see someone make a point about AI when they are misusing it and then making a statement about its actual usefulness or capabilities, it does cause an adverse reaction deep in my soul. Becuase, it is very important that people know how powerful and consequential it is to their future.
I apologize for the overreaction, but the question still remains. What was the point you were trying to make that I so poorly understood?
Are you AI?Hi Erik,
Sorry for that. What did you mean by this comment?
Addy & Arianna transferred to WVU according to AI.
I assumed you were suggesting that AI's "answer" was incorrect, proving some point you were making.
You are correct about my response, I will give you that. Every time I see someone make a point about AI when they are misusing it and then making a statement about its actual usefulness or capabilities, it does cause an adverse reaction deep in my soul. Becuase, it is very important that people know how powerful and consequential it is to their future.
I apologize for the overreaction, but the question still remains. What was the point you were trying to make that I so poorly understood?
