Denver, CO based MBA/Developer with a long history in professional 3D animation production; over 15 years as a lead game developer (Sony, EA), several years in feature film visual effects (Rhythm & Hues Studios), global patent author of what are now called DeepFakes, founder and CEO of www.3D-Avatar-Store.com (site is gone, who knows what is at that link now). Was working full time in facial recognition as a software scientist but left that industry min-Pandemic. Currently CTO of a CA Immigration Law firm, CEO/developer of https://midombot.com/b1/home and also developing propeller-head targeted courses. Contact me if interested in learning what an accomplished Software Scientist with an International Business & Economics double major MBA could do for you...
To find me for communications, search my username on Google, and you're likely to see various places I can be contacted.
I'm sure many have made something like this, I've done a few. I've found simply submitting one's prompt to multiple models to be kind of pointless. You're just going to get statistical noise from the variances in their training methods, as they are all training on pretty much the same data.
I get significantly better results by pre-prompting each LLM (they can be the same LLM too, just another instance), I pre-prompt them to approach from a different perspective. Basically, I create expert personas that each believe they are someone of a different career, different intellectual perspectives, and then that generates a real debate between experts.
Serious question: I find that the core issue in corruption, and the corruptibility of an individual and then the groups they are a member relies on plain spoken communication skills, and the ability or inability of an individual to protect themselves from self deception.
A "uncorrupt person" does not choose to be corrupt, they lie to themselves with little corruptions until "it is too late" and then with no (self assessed) alternative they commit to their situation.
As well as a group of people knowingly engaged in less than ethically honest activities must impress their ethical failures on all their peers, to insure their safety if the corruption is exposed, and they then bully their uncomprehending peers into corruption.
All of this is caused to the educational hole that is effective communications: people are not taught how to identify honest and dishonest self conversation, people are not taught how to communicate without information mistakes they refuse are mistakes, omissions of critical information, and then deferment of blame when half considered strategies are met by complex reality.
I've been working in tech, at all levels, since the 70's. The #1 characteristic of every single organization I have been a member and interacted with is poor communications, weak documentation, and nobody able to discuss the topics at hand without technical pretension, omissions, misdirections, and misinformation.
This nonsense is all due to no real emphasis on communicating to mutual understanding anywhere in the verticals of technical and science educations. And that includes the communications in people's heads that drive them to comforting non-solutions they attempt and fail.
The issue is anything that is of value requires some level of detail, of complexity, and that is only of interest to people that know that specific complexity, and it is a pain point for them. Now they'll care. Everyone else? Lost them. So, the marketing challenge is to find some aspirational complexity that people wish they knew, and how that can be solved with AI, and without turning that thing into a trivial nuisance, but a valued skill. That logical series right there is, well, too much for far too many.
When LLMs and ChatGPT first came out, it struck me as obvious and dangerous to a deep thinker or a knowledge worker the answering capacity. So, from my initial use I did not ask them questions, I have always "done my own work" and then asked the LLMs to criticize that work. This has been an exponential ladder of learning, and my cognitive growth is personally noticeable. I'm not hesitating to scribble out calculus and work it out, as I need for my work, where in the past I'd have found some other way because I felt uncomfortable with my tip-of-my-tongue calc skills. Don't ask AI, do your own work and ask for criticism, and them improve your own work yourself. This creates a learning ladder that you will climb.
The blog post this discussion is for is one of the first in depth discussions I've seen of how these coding agents work. Most posts cover how to use them, not their internals and how they operate.
There needs to be a realization of how important communication skills are to develop and possess. The act of disagreement has skill levels that do not trigger emotional responses, and cause cross understanding to occur. Learning how to convey understanding and gain understanding from others becomes more and more important in a landscape of rapid change. Which we are collectively terrible at, with most companies being miscommunication circuses, with all the stress that generates, needlessly.
Is there an AI Coding Agent application structure emerging that is more or less universal across llm models? Is anyone collecting and writing on how to understand this architectural style?
Around the same time I first read On The Road, my wiser than should be possible mother said "Oh, you need to read this too" and it was "Off the Road: Twenty Years with Cassady, Kerouac and Ginsberg", written by Carolyn Cassady. Rips the band aid right off of those sexist abuser of innocents, those utter assholes. They write great literature, and the fact that they expose their own terrible ethics bare, but surrounded by non-condemning language is the trick. They never hid their nature, but America never realized what they were praising, not really. Which is all so American!
I worked at a shop where we used Boost in a C++ code base that the only use of C++ was the harness to use Boost. After that, it was all C, object-styled C, as that code base started before C++ compilers were not a template overlay on C.
One of the first homework assignments when I learned C back in '83 was after a long lecture on how the string functions are fundamentally broken, and the class introduction to writing C was fixing all of them.
We have a distinct poverty when it comes to secondary considerations and long term ramifications - which used to be manageable when progress was slower. Now, we're on a very steep acceleration/progress curve, and any shortsighted mistakes cause extremely large ramifications. Which are then compounded by both more short sighted non-fixes and our rapid acceleration/progress curve layering in additional confusion, misunderstandings, omissions of critical information.
This is a critical communications issue that is becoming what I believe the defining characteristic of "This Age": nobody knows how to discuss disagreement, and because it cannot even be discussed communication ends, followed by blind obedience, forced bullying, retreat and abandonment. This is going to be a hell of a ride, because nobody can really discuss the situation with a rational tone.
Maybe this is relevant? I worked in animation and VFX for an Academy Award winning VFX studio and several well known animation / game studios, starting around '90. I formally left the industry around '04 to work on my own tech startup. When I left, there was a lot of R&D work surrounding the huge amounts of data that an animation studio generates and works with; I was one of those people creating early deep learning systems for production forecasting.
Anyway, right around '10 the industry was really stressed. The financial crash was 2 years in, and the recovery was more propaganda than reality. The productions were chasing a Hollywood market that the population did not have the disposable income to support. Then in all that stress, the Me-Too movement starts. Rumors and murmurs at first, but soon a tsunami of women from the entertainment industry sharing their institutional abuse and choosing to leave the industry entirely. My wife was one, an Academy Award winning filmmaker, famous for children's media.
That line in time of Hollywood films going bad? It is when the women that were silent in their abuse chose to leave the industry enmasse. What replaced them were clueless men and women okay with the abuse, and the reduced quality of Hollywood is a reflection of the quality of their intellects.
I remember, my older brother wanted to be an Eagle Scout, and I was of the age to initially join. Some respected counter culture person, I don't think I was him, but someone like George Carlin, started talking about the boy scouts being where men scouted boys. That was it, the boy scouts were instantly "uncool".
This Gen Z resentment is manufactured, so there is yet another pool of people that are angry enough to deludedly back the next aggressive idiot "savior", justifying an attack on the general population, ensuring authoritarianism is viewed to be the "only way forward."
I was bullied, a lot. The only thing that really stops a bully is shaming them, treat them like they are a child that needs to be treated as younger than their age. Bullies are demanding and demonstrating dominance, and if that is turned around on them, the bullying stops.
Bullies need to be identified as simply immature, treated as children that have not graduated to their age. That really impacts the individual. Make them wear identifying clothing as a "special case" and they will mature very fast.
This is why I avoid the entire JavaScript shitshow that is NPM and all that ecosystems nonsense. The population of users do not have the secondary considerations to be trusted, there will always be someone that does the worse and talks too many into following them. Then the "best practices" produce failures. What a shit show.
I never went to agentic loops. I've been doing "agents" since the initial release of OpenAI's API. I've also been an active AI researcher for 40 years, surviving the winters by also being a 3D graphics and animation pipeline developer.
I am approaching AI as not an automation technology, but as a cognitive assistance technology. My AI Agents are all "method actor prompted" and placed inside other open source software, where they are created to believe they are one of the developers of that open source application. That creates "intelligent" versions of those applications.
For example, a word processor I have integrated agents into that are able to support editing and layout of the contents of the editor, but more important the user can create solo and groups of agents that have knowledge of the subject document, and understand how to communicate better than the user - BUT they do not write for the user, they coach the user to write and communicate better.
Are you familiar with Chain-of-Thought / Reasoning models? I've made a fully transparent system that is similar I call "Chain of Method Actors" that operates very much like Chain-of-Thought models, but the user is included in the loops, in a conversational exchange between their experts and the user. Net result? Massive efficiency and expense savings. Where CoT models may spin in their agentic loops hundreds of times before they emit their best guess what they user wants, my Socratic Chain of Method Acting Agents will begin with clarifying questions, rather than guessing, and the eventual answer the user seeks is often delivered in 3 to 6 loops. Not hundreds.
But my goal here is not the same as most: my system I have designed is for people doing critical work they must understand and stand behind. This is a knowledge IDE that I have created. It's an office suite with conversationally programmable AI Agents that method act the expertise that one needs. It is not automation, it's human enhancement, intellectual cognitive support for people that cannot trust automation or delegation, the task is too critical.
I think the "answer" to users that anthropomorphize their LLMs is to create for them an environment where they are on both sides of the LLM: give users the ability to create their own contextually limited, purpose driven AI Agents where they control the "personality" of the agent, because what the agent is doing is context-wise something the user understands intimately (its their career) and they actually understand the problem space better than a software engineer (occupant of a different career) could and would care to deeply understand at the same level.
I've actually set up an environment like this. It requires contextually positioned agents with a limited scope and purpose. Imagine something like a creative writing agent that understands the literary genre of its user, and the user is able to change the focus of the literary agent, or create new ones that provide a counter-point perspective. As the user operates their word processor, the agent(s) can be asked for opinions, advice, and so on. But the point being: when the user is on both sides: they authored the agent, and that authoring was entirely purpose focused, nothing technical unless that what the user instructs them to understand. With the user on both sides, dangerous anthropomorphism is largely erased, their agent is only doing what they told it, and when it does something unexpected they an easily reason why. Magic AI no more.
I think he is saying, I hope he is saying, that software has never been writing software, it has been communications with people over what the software should be, needs to be, and the entire point all along has been to achieve better collaboration with people, and implied: to achieve their collective goal. He spends a good amount of time on how slow writing software has been in the past, and that allowed the industry to over focus on the software writing. While it has been pointed out a number of times by milestone books our industry embraced that it is the communications aspect of why and what we write that is the most important. Finally now that is being forced upon us because writing code is now automated, and all that is left is the specification and the communications with humans over what and why.