Do you use AI @work without formal employer approval?

If I ran a company again, I'd let them but the motto would be "Don't trust. Do verify." These AI tools are similar to search engines only. Their results may be inaccurate if they work at all.
 
Yes like abbreviations. As humans we have a hard enough time.
SLA for batteries or filler wire? DRM for content or video rendering?

No computers where I work because IT got canned. I have hacked into a MSC Tool Kiosk to do my daily time...
Some kind of SupplyPoint rig with dell minis inside.
 
I asked grok to give me a http server in tcl hat takes the request and shows it in stdout. This was the answer that immediately came. it works:

#!/usr/bin/env tclsh
# Simple echo server: prints full raw HTTP request to stdout

proc accept {chan addr port} {
fconfigure $chan -blocking 0 -buffering none -translation {binary crlf}
fileevent $chan readable [list read_request $chan]
}

proc read_request {chan} {
if {[eof $chan]} {
close $chan
return
}

# Read everything that's currently available
set data [read $chan]

# Accumulate data
append ::request_buffer($chan) $data

# Try to find the end of headers
set headers_end [string first "\r\n\r\n" $::request_buffer($chan)]

if {$headers_end == -1} {
# Headers not complete yet → wait for more data
return
}

# We have at least the headers
set headers [string range $::request_buffer($chan) 0 $headers_end+3]
set remaining [string range $::request_buffer($chan) $headers_end+4 end]

# Look for Content-Length
set cl 0
if {[regexp -line {^Content-Length:\s*(\d+)} $headers -> length]} {
set cl [scan $length %d]
}

# Do we have the full body yet?
if {[string length $remaining] >= $cl} {
# Yes — we have the complete request
set full_request $::request_buffer($chan)

# Print the entire raw request to stdout
puts "----------------------------------------"
puts "New request from [chan configure $chan -peername]"
puts $full_request
puts "----------------------------------------"
flush stdout

# Optional: log just the first line + content length
set firstline [lindex [split $headers \n] 0]
puts stderr "→ $firstline (body: $cl bytes)"

# Send minimal response
puts $chan "HTTP/1.1 200 OK\r\n"
puts $chan "Content-Type: text/plain\r\n"
puts $chan "Connection: close\r\n"
puts $chan "\r\n"
puts $chan "Request received and logged to stdout.\r\n"
flush $chan

close $chan
unset ::request_buffer($chan)
}
# else: body not complete yet → wait for more data in next readable event
}

# Main
puts "Starting raw HTTP request dumper on port 8080..."
puts "Send any HTTP request (GET/POST/PUT/...) — full raw content will appear on stdout"

socket -server accept 8080
vwait forever
Click to expand...

Did it really create it, or took it from somewhere?
I am seeing results, speaking with KI, and still skeptic!!!!
 
We are "allowed" (I think) to ask very generic questions (probably in the spirit of "what is the temp in degF for X degC?") now but, no we can't generally use AI; even for meeting minutes or any sort of "general type office paperwork".

I really don't have time to read up on the whole AI thing at work just because there is so much legal background/involvement/aspects so I typically just avoid it or use it for stupid things like "converting feet to meters" if at all. It's generally scary enough just wording emails, involving third parties, by myself (adding anything else into the mix would be just plain stupid!). I'm really not sure what (little/low liability wise) tasks I would use it for.
 
If I'work in a place when pay me too low,suffer from workplace abuse,had an asshole boss
then yes, I use AI for everything,even let the AI manages the rrhh and employee salary payments and deposits

for other side,if I like my work and care take of my servers is a big NO
fuc# AI
 
I asked ChatGPT to write a brief compelling argument against the use of AI.

ChatGPT:
Every time you outsource thinking to a machine, you quietly trade away the friction where judgment, responsibility, and originality are forged, letting convenience hollow out the very skills that make you human. A world that hands its curiosity, memory, and moral weight to systems that do not suffer consequences is not advancing—it is rehearsing its own irrelevance.
 
atax1a said:
we used to write tcl professionally and this is terrible code that we would hate to debug
Click to expand...
grok may help you to debug it ...

But my question is not answered: did it really create it?

Look at the message:

puts "Starting raw HTTP request dumper on port 8080..."
Click to expand...

grok called it 'request dumper', I did not use that word, just described in very few words what it should do.
 
hruodr said:
But my question is not answered: did it really create it?
Click to expand...
Yes and no. It's most likely not a carbon copy but it will be stuff from inside the training data. The llms most likely roam inside the envelope but can't leave it - or has anyone seen something different?
 
of course it didn't create it. it can't create anything, it can only, using a discredited theory of language and intelligence, replicate something that resembles statistically-likely output.
 
You must log in or register to reply here.
Back
Top