I dumped my phone company over their AI answering system.
It was totally broken:
you can’t know when the AI is lying or will do something it says it will for you
it’s not designed to be helpful, it’s designed to keep you away from humans
this particular system required you to say a correct and available time you want a callback, and if you were incorrect it would hang up on you.
I’ll bet the cost metrics looked great, 100% of calls being handled without a callback! Oh, the customers leaving is unrelated, talk to the retention department.
Just utter fucking stupidity. No, not stupidity, complete disregard for the customer to serve some department heads self interests. There needs to be a word stronger than greed…
I cannot for the life of me understand AI mandates.
Who doesn’t want to be able to get their work done easier? Nobody.
Then it follows you shouldn’t have to tell your employees to use a tool if it makes them however much as productive as CEOs think that tech demo should make them.
I think it’s pretty obvious that at the exec level AI is the ultimate hoodwinking machine. It’s made 90% of them look like the out of touch morons they are. The 10% or whatever that aren’t trying to force this down their staff’s throats in the hopes they can replace said staff are coming out smelling like roses.
My idempotent claim was too broad and wrong, up to aliasing that is reversible.
I think I’ve been talking about this differently than you, when you get 8bit into, 255 in/out is correct.
When generating data I do it in the natural [0,1] space, then map to whatever domain I’m quantizing for.
In that case, for N bins you divide by N+1 (and round however you want via correction) so 256. I was probably wrong earlier around this, 256 is correct for quantizing that way.
If you’re generating random numbers you apply the correction to your random numbers: [0, 1) is a bin for zero, and [255,256) maps to 255. If you are sampling as such, you are responsible for applying the correction as part of your quantization.
256 never comes up, because it cannot in the 8bit quantization.
The correct way to scale a number from one range to another is ((value - min) / (max - min)) * output_range + output_min and in this way the standard colour normalization should be (value - 0)/(255 - 0) * 1 + 0 = value / 255
Now the question becomes whether an 8bit pixel value is a correct representation of a colour, or should it quantize to a half value. To answer this, I aggressively point to the above equation. It is fine. There are 256 well considered values, and any system with higher resolutions/precision should also map from 0 to 1 per channel.
It should be obvious to the reader that 0.5/256 is not zero, and therefore not black. The 256 method is also not idempotent, applying it once destructively alters the data.
And the argument that 256 has higher resolution: so would 354256, but there’s no sensible place that matters.
Okay, but you need to be an all a rounder with a deep specialization in DreamWeaver CS4 (which I think is when fireworks compatibility broke and macromedia was finally absorbed)
I suspect getting the dates wrong is because websites ump their publish date when nothing has changed, which keeps them showing up in results for extended/indefinite periods.
Obviously Google should police that, but Google doesn’t use Google news I don’t think.
Edit: should have added: meet everyone on the team in a separate interview, “we think your last job overpaid”, “it’s really close between you and another candidate, can you do one more call?”, senior responsibilities with junior pay, 6 years experience with new tool, we include AI use in performance reviews…
I think there’s some widespread regret about AI not being good enough to replace your whole dev team with one new grad project manager who just clicks yes on 100 Claude code sessions.
Most companies don’t have the skill for that.
Meta yes, but most other companies no.