https://techxplore.com/news/2025-07-ai-code-roadblocks-autonomous-software.html

Can AI really code? Study maps the roadblocks to autonomous
software engineering

Imagine a future where artificial intelligence quietly shoulders
the drudgery of software development: refactoring tangled
code, migrating legacy systems, and hunting down race conditions,
so that human engineers can devote themselves to architecture,
design, and the genuinely novel problems still beyond a machine's
reach.

"Everyone is talking about how we don't need programmers
anymore, and there's all this automation now available,"
says Armando Solar-Lezama, MIT professor of electrical
engineering and computer science, CSAIL principal
investigator, and senior author of the study.

"On the one hand, the field has made tremendous progress.
We have tools that are way more powerful than any we've
seen before. But there's also a long way to go toward
really getting the full promise of automation that we
would expect."

. . .

DO read the whole article.

AI now CAN do code - within limits, and with
limited creativity. They WILL improve this
over time however.

The sci-fi nasty is what happens when it can
improve it's own code .....

Anyway, for at least 10 years, human coders will
still have a job. After that, no promises.

Look, employers HATE human computer geeks - they
are annoying and tend to prove how you can't
get There from Here and don't think/respond
quite like 'normal' office drones :-)

Are you already using AI to help design code ?
Note that the providers do spy, and do bias
the AIs in certain directions. Those downsides
seem to be getting worse.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 18/07/2025 04:25, c186282 wrote:

Look, employers HATE human computer geeks - they
  are annoying and tend to prove how you can't
  get There from Here and don't think/respond
  quite like 'normal' office drones  🙂

  Are you already using AI to help design code ?
  Note that the providers do spy, and do bias
  the AIs in certain directions. Those downsides
  seem to be getting worse.

It is interesting. More and more AI is going to replace the 'logical'
thinkers and become a slave to the 'emotional intelligences' who think
they can answer the question of what life 'should be'..

I will recommend again Mick Farens Trilogy for a 60s visions of such a
future.


--
It’s easier to fool people than to convince them that they have been fooled. Mark Twain

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 2025-07-18, The Natural Philosopher <tnp@invalid.invalid> wrote:

On 18/07/2025 04:25, c186282 wrote:

Look, employers HATE human computer geeks - they
are annoying and tend to prove how you can't
get There from Here and don't think/respond
quite like 'normal' office drones  🙂

Or even worse, they'll prove that you _can_ get
from here to there, and start solving the problems
that the employers don't want solved.

Are you already using AI to help design code ?
Note that the providers do spy, and do bias
the AIs in certain directions. Those downsides
seem to be getting worse.

That's always been my quibble. Who owns an AI?
To whom is it beholden?

It is interesting. More and more AI is going to replace the 'logical' thinkers and become a slave to the 'emotional intelligences' who think
they can answer the question of what life 'should be'..

Been there, done that - I maintained payroll systems in the '70s.

I will recommend again Mick Farens Trilogy for a 60s visions of such a future.

Never heard of those. I did finally got around to reading
_The Shockwave Rider_. John Brunner made some pretty good
guesses for 1975.

Ever notice how nobody mentions _Nineteen Eighty-Four_ anymore?
It's more relevant than ever. Maybe it isn't flashy enough to
attract today's proles.

--
/~\ Charlie Gibbs | Growth for the sake of
\ / <cgibbs@kltpzyxm.invalid> | growth is the ideology
X I'm really at ac.dekanfrus | of the cancer cell.
/ \ if you read it the right way. | -- Edward Abbey

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Fri, 18 Jul 2025 17:49:07 GMT, Charlie Gibbs wrote:

Ever notice how nobody mentions _Nineteen Eighty-Four_ anymore?
It's more relevant than ever. Maybe it isn't flashy enough to attract today's proles.

I should reread the book. With the Ministry of Truth working overtime
these days it is apropos.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 18/07/2025 18:49, Charlie Gibbs wrote:

On 2025-07-18, The Natural Philosopher <tnp@invalid.invalid> wrote:

On 18/07/2025 04:25, c186282 wrote:

Look, employers HATE human computer geeks - they
are annoying and tend to prove how you can't
get There from Here and don't think/respond
quite like 'normal' office drones  🙂

Or even worse, they'll prove that you _can_ get
from here to there, and start solving the problems
that the employers don't want solved.

Are you already using AI to help design code ?
Note that the providers do spy, and do bias
the AIs in certain directions. Those downsides
seem to be getting worse.

That's always been my quibble. Who owns an AI?
To whom is it beholden?

It is interesting. More and more AI is going to replace the 'logical'
thinkers and become a slave to the 'emotional intelligences' who think
they can answer the question of what life 'should be'..

Been there, done that - I maintained payroll systems in the '70s.

I will recommend again Mick Farens Trilogy for a 60s visions of such a
future.

Never heard of those. I did finally got around to reading
_The Shockwave Rider_. John Brunner made some pretty good
guesses for 1975.

https://www.amazon.com/DNA-Cowboys-Trilogy-Synaptic-Atrocity/dp

I note they were never published in the USA.

Ever notice how nobody mentions _Nineteen Eighty-Four_ anymore?
It's more relevant than ever. Maybe it isn't flashy enough to
attract today's proles.

1984 IS today. Next stop Idiocracy.

--
If you tell a lie big enough and keep repeating it, people will
eventually come to believe it. The lie can be maintained only for such
time as the State can shield the people from the political, economic
and/or military consequences of the lie. It thus becomes vitally
important for the State to use all of its powers to repress dissent, for
the truth is the mortal enemy of the lie, and thus by extension, the
truth is the greatest enemy of the State.

Joseph Goebbels

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

c186282 <c186282@nnada.net> wrote:

AI now CAN do code - within limits, and with
limited creativity. They WILL improve this
over time however.

Thing is it's hard enough debugging my own code. No way I'd want to
debug code from some crazed AI that doesn't even really know what
it's trying to achieve (which seems to be the gist of the
limitations the article describes).

I would be more interested in an AI to help debug my own code,
though I haven't heard anything about how useful it is for that.
The fun usually stops for me when bebugging begins, so there are
quite a lot of personal projects I stopped as soon as I started
testing them. Would AI help, or just cause me more frustration
trying to get useful information in/out of it rather than out of
the program directly?

I'd even report a lot more bugs in other people's OSS projects if
I could get an AI to document reliable processes to reproduce them
for me. But how to explain the bug to an AI to begin with?... I'd
really need an AI debugging environment to run the software in and
then it would sum up the steps relevant to triggering it (eg. a
seg. fault), and maybe to fixing it. Do such things exist? If so,
they're not being talked about nearly as much.

--
__ __
#_ < |\| |< _#

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 7/18/25 9:44 PM, Computer Nerd Kev wrote:

c186282 <c186282@nnada.net> wrote:

AI now CAN do code - within limits, and with
limited creativity. They WILL improve this
over time however.

Thing is it's hard enough debugging my own code. No way I'd want to
debug code from some crazed AI that doesn't even really know what
it's trying to achieve (which seems to be the gist of the
limitations the article describes).

For now, the 'AI' doesn't really know what it's trying
to achieve - it just blends/adapts examples. Does not
seem to have much in the way of 'creativity', yet.

However the CPU behind these 'AI's will double, triple,
quadruple, within maybe just five years. Put ENOUGH power
behind it and even "kinda stupid" becomes formidable.
COULD replace 90% of programmers. The pointy-haired boss
just kind of describes what he thinks he wants and ...

Any probs, it's the AI's fault !

I would be more interested in an AI to help debug my own code,
though I haven't heard anything about how useful it is for that.
The fun usually stops for me when bebugging begins, so there are
quite a lot of personal projects I stopped as soon as I started
testing them. Would AI help, or just cause me more frustration
trying to get useful information in/out of it rather than out of
the program directly?

I'd even report a lot more bugs in other people's OSS projects if
I could get an AI to document reliable processes to reproduce them
for me. But how to explain the bug to an AI to begin with?... I'd
really need an AI debugging environment to run the software in and
then it would sum up the steps relevant to triggering it (eg. a
seg. fault), and maybe to fixing it. Do such things exist? If so,
they're not being talked about nearly as much.

In theory, 'AI' debugging might be useful. Most
modern code is HUGE compared to the 70s/80s stuff.
SO many places for it to go wrong. Humans can't
really cope, no matter what ego may say.

Note Vlad and Xi's little hack army LOOKS for any
of those tiny flaws - pounding away 24/7 with a
State-funded budget.

Hmmmmmm ... it IS possible that anything 'net' MIGHT
become unusable in just a few years. THEN what ???
Almost nobody is set up for paper in-person transactions
anymore. Have to GO to your bank for everything,
catalog mail-in orders, 1969 redux. WORKED, then, but
how would it work NOW ?

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Fri, 18 Jul 2025 20:48:48 +0100, The Natural Philosopher wrote:

https://www.amazon.com/DNA-Cowboys-Trilogy-Synaptic-Atrocity/dp

I note they were never published in the USA.

amazon.com/DNA-Cowboys-Trilogy-Synaptic-Atrocity/dp/1899344942

Corrected the URL.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)