Pondering shutting off that precocious Apple Intelligence toddler I’ve loosed into my day, it being all too prone to re-sorting and re-organizing and re-prioritizing.
I do like the summarizing tool, but only where I can ignore when the summary diverges from the original; where the summary is inadequate, inaccurate, or just wrong. Where the results don’t matter.
Where the results do matter, AI LLMs are too often enormous “Where’s Waldo?” quests for the errors.
So LED lighting is commonly characterized as warm, neutral, cool, and cold. Annoyingly.
This is measured using color temperature.
Roughly…
2000ºK to 2700ºK for warm lighting,
3000ºK to 4000ºK for neutral,
5000ºK to 8000ºK for cool and cold.
Why annoying?
The chosen color names run completely opposite the measured temperatures.
What’s called the coldest color temperature lighting is actually far and away the hottest.
🤷🏼♀️
Selecting the appropriate color temperature for your home's lighting is a crucial choice that can impact your living area's atmosphere, mood, and usability. The various temperature options can produce diverse effects, ranging from a warm and comforting ambiance to a cool and invigorating one. To bring out the best in…
AI hallucinations are inevitable.
Duh.
Next week’s revelation?
Not just “Industry evaluation methods made the problem worse”, but industry _valuation_ methods made the problem worse.
Next up, studies in self-deception, mass deception, advertising, hype, propaganda, and believing your own and ill-considered bullshit.
In a landmark study, OpenAI researchers reveal that large language models will always produce plausible but false outputs, even with perfect data, due to fundamental statistical and computational limits.
Now we get to answer friends’ and relatives support questions, and explain why some blithering stochastic parrot AI is blithering.
Garbage-Producing Tools.
LLM and ML are useful for some cases, but other cases are too often flaming hot ripe garbage.
Stephen Hoffman
George Pearkes
realcaseyrollins ✝️
13Rx