I think that a lot of the accounting, used mostly in factories and such, is having trouble with the new financial reality in the tech space. Running a factory, having a debt to asset ratio over 1.0 would be disastrous because the assets are unable to produce revenue fast enough to cover the interest.
With tech it's weird because something that is worth nothing on an accounting sheet could be worth a hundred billion dollars during a sale to Alphabet or Meta.
Yeah, sometimes you just have to throw out the current solution and start over/rollback. AI will go down some weird rabbit holes to solve a problem with the wrong requirements. Sometimes I feel like I'm just coding in English instead of Python or TypeScript.
If I'm working on a non-coercive environment (No deadlines, requirements, etc..) hand coding in hyper-focus mode is extremely zen. But working on a deadline it really hurts. My brain can't help but ratchet up the stress.
I agree, the speed at which the agents work is a problem. It creates an interrupt workflow instead of a flow workflow. This works for some people but I notice that I don't do well that way. Typically I'll have an agent working on a task while I use a separate one in Ask mode just kind of satisfying my curiosity while I wait. This keeps me engaged but makes the interrupts less jarring.
I've found that my requirements and acceptance criteria are much less clear than I think they are. AI reveals when you are being too vague, almost instantly sometimes.
I've noticed that you can either move very fast with AI, or very slow. Moving fast, you place a lot of trust in the agent and let it proceed and create something very quickly that works at a surface level and that you don't understand at all. Moving slow you take the time to ask questions and understand every part of the application and learn more about how to code it yourself and make changes as necessary.
One thing I've noticed is, that coding manually is exponentially more stressful than the slow AI approach. I'm not going any faster than I normally would, but the burnout doesn't show up on the fifth day of the week. The hyper-focus is kind of fun but utterly exhausting.
With the fast approach I can create impressive demos and POCs that are nowhere near ready for production. But sometimes that's exactly what I need.
The problem is that environmental regulations kind of act like a double edge sword with regards to hybrid conversions. They prevent Jim Bob from selling his environmental disaster kit, but they also prevent Bob Jim from selling his Hybrid drivetrain conversion without expensive approvals. This is why most people do BEV conversions. The kits are available because they are zero emissions and don't fall afoul of the environmental regulations.
Welcome to the Kafkaesque world of automated enforcement.
> "The Trial" by Franz Kafka is a novel written in 1914 and 1915. It follows Josef K., a bank clerk who is arrested and prosecuted by a mysterious, unreachable authority. Neither he nor the reader ever learns what crime he has committed. As Josef navigates an absurd legal system filled with bizarre encounters and inexplicable procedures, his case consumes his life while remaining perpetually unresolved. The novel, never completed by Kafka, was published posthumously in 1925.
Ecologically, wildfires are necessary for some biomes. However, not all of them. Many are needlessly destructive due to climate change and over logging of forests.
This is honestly kind of accurate accidentally. The Luddites were a movement trying to stop the exploitation of labor to work in dangerous factories which was destroying the health of large groups of people for the sole benefit of factory owners. The underlying problem isn't the AI, or the factories. It's the anti-social actions of rich sociopaths.
A lack of trust in corporations and the ultra-wealthy who own these models is only natural. Especially considering the use of AI has so far been less than ethical. (Targeting decisions for air strikes, surveillance, blatant IP theft, fraud, etc...)
I think the biggest advantage rails have is that they are electrified and are accessible by rail based maintenance vehicles. Panels have gotten so inexpensive that the major cost is usually the mounting, installation, and electrical access. Some of that is mitigated by installing on railways. You can automate the installation process and connect into the tram lines.
If you save a child from malaria but they die in a civil war a few years later you have saved one life (Temporarily) but had very little impact on the death rate. Using changes in death rates to gauge the actual number of "lives saved" from preventable illness is going to be fraught with confounding factors.
USAID certainly had a massive impact in mitigating death from starvation and illness. They also did a great deal in spreading goodwill for the USA among nations that would otherwise have only seen us as only a shadowy menace thousands of miles away.
LLMs are essentially a text prediction engine. This can be used for basic reasoning tasks, however the LLM doesn't have much in the way of actual knowledge. However, it can use knowledge that exists to predict text better. For instance, if you plug a bunch of scientific articles into it, it will be really good at answering questions about the subject of those articles.
The problem is that context windows are very short. 200k-1M tokens or so. This means that the model needs to focus down on very specific information if possible. This is what makes tool using, reasoning, and agentic AI very powerful. The model can find the most relevant information it needs within its limited context and generate relevant answers to questions. The LLM pulls from web searches, documentation, long term memories in graph databases, and database queries to answer the questions using real information.
Raw dog Chat LLMs are pretty worthless. But run an agent with tool invocation and they get scary good. It's amazing how much reasoning is packed into the English language. Provide your model with enough information and it can pull some miracles out of thin air. It's not the "Replace humans" level yet, but you can automate a lot of stuff you wouldn't expect to be able to automate.
I think that this is doable. Similar to having a new employee shadow a more experienced one and observe, you could implement a sort of program where AI shadows experienced employees and asks questions when they do something it doesn't understand.
But this is difficult to implement since AI doesn't have a body to follow someone around and it would take immense amounts of compute to do so using telemetry and cameras. You would literally be spying on employees 24x7 for weeks at a time with the express goal of replacing them someday.
I think most of them were losses by attrition. Where they don't replace lost employees. That's usually the preferred method of downsizing if you can get away with it.
With tech it's weird because something that is worth nothing on an accounting sheet could be worth a hundred billion dollars during a sale to Alphabet or Meta.