Writes code.
Realises that debugging code that was written by the lunatic that is yourself two nights ago is going to be a big part of her life.
dies
We've all had debugging sessions where that feels like the best option. Right?
Writes code.
Realises that debugging code that was written by the lunatic that is yourself two nights ago is going to be a big part of her life.
dies
We've all had debugging sessions where that feels like the best option. Right?
given how mechanical it was, it likely wouldn't be hard to colour some of the memory gears and make it so they are visible from the outside turning it into a display.
Pff, who needs a display? Just do that Matrix thing and render the raw state in your head.
But to the user, it can correct their inputs. The rest is an abstraction. My point is that there’s more to a platform than just precise calculations. Obviously the asker isn’t thinking this far ahead, but Babbage is also rather flippant in his response.
That quote is my favorite example of a very polite wtf
Debugging was easier when all you had to do was spray the room with fly spray and vacuum the tubes.
This I do not know. However the first widely reported railway fatality in the UK (possibly the world) was MP William Huskisson who was run over by Stevenson Rocket...
Not relevant to the conversation but I think it's interesting.
So the I-d10t bug has been around since the beginning, it seems.
it has limited memory but could me expanded, by a lot. but I'm theory yes, no display though.
Being extremely, extremely generous, maybe they meant a human would notice the input was incorrect? But even then, a human could notice the same when inputting it into a computer.
Not exactly. An autocorrect is a closest-match or prediction device with correct input given beforehand. When you type "fridsy", what it does is to answer the question "between fridsy and this set of words, what is the shortest distance?"
Works on my machine
I've seen a couple papers on theoretical designs for purely mechanical computers that can run doom, but as far as I am aware I've never heard of one that's actually been built.
but in theory yes it could have
Could it run Doom though?
One time my boss asked me to basically solve the Travelling salesman problem.
My first pass at ot was a simple grab closest neighbor solution, but that left a slightly unoptimal path and my boss asked me to "fix" it. I explained to him why, no, I can't make it both fast amd accurate, pick one, while also showing him that wikipedia page. I was so mad when he said just make it more accurate ignoring now it takes hours to run sometimes only to save 10 seconds of a machine moving.
Clients? Shit happens in my house.
"My monitor keeps turning off."
"Ok next time it happens ill look at it and see if i can figure out what is going on."
"Cant you just fix it?"
"Fix what? I dont know whats wrong yet."
"Just fix the monitor."
Isn’t the whole point of autocorrect to put the wrong input but still get the right output?
This is how I expect AI to work. I will silently think of a thing, and the AI must make it perfectly in one go. If it doesn't, I have just lacked in describing in detail what it should do. And that takes thousands of lines of code.
"If I fuck up, will it correct it?"
"No."
"Will ChatGPT correct it?"
"Yes. Too much."
All unit tests show PEBKAC
Big ones.
"Can it ChatGPT?"
"No."
"Can ChatGPT?"
"No."
No idea, but wouldn't be surprised at all.
Don't you mean 5 and a quarter pence?