Raw LLM Responses

[The cost of AI is dropping like a rock.](https://youtu.be/T17bpGItqXw?si=uGJ-NYdcAivMKg2g) The cost of self-hosting an open-source model like Deepseek or Mistral that is capable of running agents for personal use or for a small business is extremely affordable and could probably pay for itself within a few years through savings on meal planning, inventory management, environmental monitoring (power and water usage, air and water quality, etc). The cloud model is still a viable model at the moment. Most people don't need an AI running 24/7, but will only use it for the few heavy computational tasks that likely only take a few hours every month. Will cloud services be as profitable as the hype? I seriously doubt it, yet the breakeven point gets lower, so profits will be there. For organizations that would benefit from 24/7 AI, i.e., universities, hospitals, municipal governments, etc, the barrier is not hardware, but useful software tools and qualified IT staff to run the systems, yet IT departments are competing with a dozen other operating functions and face severe budget constraints as well. The lower cost of hardware helps the business case, but may not be enough to offset the other constraints. I see the largest barriers being the same inertia and indifference that drives all technology adoption. ERP and RPA systems are decades old, yet adoption rates are still fairly minimal with about 50% of companies using ERP and about 20% using RPA. The percentage using them effectively and using their full capabilities is probably drastically lower. (My own experience with Accounting Information Systems follows that trend as well. Most people still use Excel as glorified graph paper for building tables and the simplest of charts. Maybe 1 in a hundred know how to use the advanced tools like statistical analysis or Power Query.) AI tools might increase those adoption rates, but I doubt more than a few points at the margin. I believe AI will become as ubiquitous as the