GIGO (garbage in, garbage out), likely coined in the late 1950’s is just as relevant today.
You need to know how and understand to operate a handsaw to use a bandsaw or a mill. You need to understand variables, loops, strings before writing an “app” (and then you need to understand user models, security models, databases, UI / UX).
You can’t “trust” AI / LLMs without first giving it fresh and relevant content -or it will make information up. You need to give it the important information for it to synthesize correctly -and that relevant info needs to be in the current context. You can’t just ask a question and expect it to “do the research”. You need domain expertise to be able to validate the response.
AI is just a tool to make it easier to use your existing tools.
:) It’s worse than that.
Part of the reason for all these new huge models to to replace the senior staff decision making process
CFO small company (25-100 employees): Whether to take the large new contract requiring hiring 15 people in 60 days -weighing cash flow risk against growth, and whether the pace breaks the culture that’s kept turnover low.
Teacher: Whether a student’s escalating behavior reflects a home crisis, an undiagnosed learning disability, or a discipline problem -each read would send the family down a very different path (and this is assuming the financial ability to do this even exists).
Non-profit program director: Whether to accept a large restricted grant that pulls staff off the program that’s actually working, because the funder’s priorities don’t match what the community needs.
Psychiatrist: Whether worsening symptoms mean treatment -resistant depression needing a medication change, an unaddressed life stressor, or an early sign of bipolar disorder misread as unipolar -each path carries different risk if you’re wrong.
Each of those examples are real life questions that are actively being worked on. It gets worse as those things I listed are “thoughtful, human concerns” that will be solved by AI with no repercussions to the corporations that utilize them. There’s no human responsibility, just a capital responsibility.
…When they are “solved”, that’s the end of the global middle class (US $150k-$1M annual salary).