OpenAI, the company behind the chatbot ChatGPT, has ramped up its hiring around the world, bringing on roughly 1,000 remote contractors over the past six months in regions like Latin America and Eastern Europe, according to people familiar with the matter:
About 60% of the contractors were hired to do what's called "data labeling" — creating massive sets of images, audio clips, and other information that can then be used to train artificial intelligence tools or autonomous vehicles.
The other 40% are computer programmers who are creating data for OpenAI's models to learn software engineering tasks. OpenAI's existing Codex product, launched in Aug. 2021, is designed to translate natural language into code.
[...] Previously, OpenAI trained its models on code scraped from GitHub, a repository site owned by its largest investor, Microsoft, which last week confirmed multi billion dollars in new funding first reported by Semafor. But in this case, OpenAI appears to be building a dataset that includes not just lines of code, but also the human explanations behind them written in natural language.
[...] Sam Altman, OpenAI's CEO, recently put the company's headcount at 375 people, a tiny number compared to the thousands of staff at tech giants like Google and Facebook working on artificial intelligence. "I know I'm not supposed to brag about OpenAI," he tweeted, touting the company's "talent density."
Originally spotted on The Eponymous Pickle.
Previously: Why OpenAI's Codex Won't Replace Coders
Related: OpenAI and Microsoft Announce Extended, Multi-Billion-Dollar Partnership
Related Stories
Why OpenAI’s Codex Won’t Replace Coders:
This summer, the artificial intelligence company OpenAI released Codex, a new system that automatically writes software code using only simple prompts written in plain language. Codex is based on GPT-3, a revolutionary deep learning platform that OpenAI trained on nearly all publicly available written text on the Internet through 2019.
As an early Beta tester, I've had extensive opportunities to put both GPT-3 and Codex through their paces. The most frequent question I'm asked about Codex is "Will this replace human programmers?" With world powers like the United States investing billions into training new software developers, it's natural to worry that all the effort and money could be for naught.
If you're a software developer yourself—or your company has spent tons of money hiring them—you can breathe easy. Codex won't replace human developers any time soon, though it may make them far more powerful, efficient, and focused.
Why isn't Codex an existential threat to human developers? Years ago, I worked with a high-level (and highly compensated) data scientist and software developer from a major American consulting firm on a government database project. Our task was to understand how a state agency was using its database to assign grants to organizations, and then to advise the agency on how to improve the database.
On Monday, AI tech darling OpenAI announced that it received a "multi-year, multi-billion dollar investment" from Microsoft, following previous investments in 2019 and 2021. While the two companies have not officially announced a dollar amount on the deal, the news follows rumors of a $10 billion investment that emerged two weeks ago.
[...] "The past three years of our partnership have been great," said Sam Altman, CEO of OpenAI, in a Microsoft news release. "Microsoft shares our values and we are excited to continue our independent research and work toward creating advanced AI that benefits everyone."
In particular, the two companies say they will work on supercomputing at scale to accelerate OpenAI's research, integrating OpenAI's technology into more Microsoft products and "digital experiences" and keeping Microsoft as OpenAI's exclusive cloud provider with Azure. "OpenAI has used this infrastructure to train its breakthrough models, which are now deployed in Azure to power category-defining AI products like GitHub Copilot, DALL·E 2, and ChatGPT," wrote Microsoft.
Related:
Microsoft Announces 10,000 Layoffs, 5% of its Workforce
(Score: 5, Interesting) by PiMuNu on Thursday February 09 2023, @01:26PM (36 children)
> OpenAI's existing Codex product, launched in Aug. 2021, is designed to translate natural language into code.
This is not possible. Natural language does not adequately describe the problem domain. Information is missing and no amount of AI can fix that.
(Score: 3, Interesting) by JoeMerchant on Thursday February 09 2023, @01:28PM (3 children)
They're not even to that limit yet.
I've asked ChatGPT basic API questions, and it's pretty good, but about 1/4 queries it will give an example that doesn't even compile.
Combine ChatGPT with an AlphaGo kind of game play engine (successful compile and test wins) and you've got something.
🌻🌻 [google.com]
(Score: 1, Interesting) by Anonymous Coward on Thursday February 09 2023, @01:41PM (2 children)
> ... it's pretty good, but about 1/4 queries it will give an example that doesn't even compile.
Interesting. I wonder, does this lead to a split in the future where "easy" programs proliferate because they can be "AI" written for next to nothing? Meanwhile, fewer "hard" programming projects get funded because they are perceived to cost too much--using the cheap stuff as a benchmark for what things "should" cost.
This had certainly happened with other products over the years--the cheap, high volume stuff drives the nice-but=expensive stuff out of the market. Take shoes for example, as a kid (c.1960) there were very fine size gradations available to find something that really fit, and the odd foot bump that I had could be accommodated by stretching the leather right in the shoe store. These days you are hard pressed to find anything outside the standard "medium" width--looking really hard you might find "wide" and "narrow".
(Score: 5, Insightful) by JoeMerchant on Thursday February 09 2023, @01:50PM (1 child)
I skimmed past an article headline where some guy got ChatGPT to write a WordPress plugin for him. If you need something that's basically "been done" many times before, and documented on the web (if you've got the patients to dig it out.)
Hmmm... makes me wonder, I need a Gnome extension...
Also, makes me think - these stories about Bing, Google, et al "upgrading" their searches using the various AI engines - are they also going to "downgrade" their basic search capabilities such that it becomes harder for us plebes to find the original content that feeds the AI engines?
🌻🌻 [google.com]
(Score: 0) by Anonymous Coward on Thursday February 09 2023, @08:50PM
I think the regurgitated material will always be bland corporate "both sides" mush.
(Score: 5, Interesting) by acid andy on Thursday February 09 2023, @01:30PM (6 children)
You're right. If they've got any sense they'll be feeding it pseudocode.
If a cat has kittens, does a rat have rittens, a bat bittens and a mat mittens?
(Score: 3, Insightful) by JoeMerchant on Friday February 10 2023, @12:08AM (5 children)
The problem isn't regurgitating the most efficient algorithms or other well known solutions to well known problems. The problem is when the customer truly wants something new / unique / never really been done before.
AI writing WordPress plugins is basically doing what we all do: searching on Google and copy-pasting from Stack Exchange or wherever.
🌻🌻 [google.com]
(Score: 3, Insightful) by stormreaver on Friday February 10 2023, @02:00AM (4 children)
I write personal property and real estate tax software, and all natural language models (I refuse to call them "AI", because they're not) will fail miserably at the kind of simple instructions I typically get:
"This taxpayer should have been exempt from taxes for 2018 and 2020, but was charged taxes for those years. Make sure he is exempt for those years."
The absolutely best case scenario is that the natural language model will hard-code those requirements into code it generates (assuming it can even make heads of tails of the requirements), which is the absolutely worst possible approach. It will also be completely ignorant of important context and resources necessary to create anything even remotely close to usable code, and will therefore generate code that will be absolutely, positively useless. In this example, the correct approach is just a couple simple database entries. No natural language model will be able to figure this out, because the solution requires a level of intelligent understanding that is anathema to natural language models.
Your natural response might be something along the lines of, "but the AI will have written the program and generated the database, so it will know all this." But it won't, because it's completely incapable of doing so without extensive work by a human programmer.
So no, natural language models will only ever be superficially helpful to real programmers (think code-completion), but will absolutely never be able to replace one. It just isn't ever going to happen.
(Score: 4, Interesting) by Common Joe on Friday February 10 2023, @03:08AM (2 children)
Although I know very little about taxes and don't work with that side of programming, I find your example excellent and "quite typical" for a program request.
Which brings me to my point: OpenAI hired 1000 contractors?! I mean, they cannot be ignorant of what you just so easily explained. Can they? I'm at a loss for words how costly this project is without directly addressing your extremely basic concern. If I were a boss with buttloads of money and thought we had a chance of developing this, I would bring in a handful of developers and instruct them to deal with this big freakin' issue. I would want a back-of-the-napkin explanation as to how they would overcome this. If those handful of developers cannot do that, I would not hire the other 995 people and waste such an insane amount of money.
I do not understand the modern business world at all.
On a personal level, I have sort of the opposite problem. I have a meeting today -- a meeting in which I've been trying to sit my boss down for over a year. I'm going to have to explain basic math to him and show him how we are only X number of people on the team and he wants X * 50 amount of work. (I will do this by explaining man-hours of work to him.) I will also have to explain why 100 tasks cannot be simply labeled high priority and expect us to magically deduce how to order these 100 tasks so we know where to start. I would think this is basic management 101 stuff, but apparently modern managers operate on a plane beyond me. Whatever they are smoking, I do not want it.
(Score: 2) by JoeMerchant on Friday February 10 2023, @10:03PM (1 child)
>I will also have to explain why 100 tasks cannot be simply labeled high priority and expect us to magically deduce how to order these 100 tasks so we know where to start. I would think this is basic management 101 stuff, but apparently modern managers operate on a plane beyond me.
I tell myself that willful ignorance is major factor in this management style. They don't want to understand reality, only results.
🌻🌻 [google.com]
(Score: 0) by Anonymous Coward on Saturday February 11 2023, @02:42AM
> I will also have to explain why 100 tasks cannot be simply labeled high priority...
Oh yeah, this happens with contracted services too. My tiny engineering company reported to a manager like that at one of the big companies. Everything he wanted from my little company was top priority, so I wound up prioritizing things for my engineer-analyst who actually did the work. I took the heat from that disorganized manager so that my good guy could focus long enough to complete one task at a time.
If I'd knuckled under to the wishes of that manager, we would have been task switching daily or more frequently (so all the tasks got done at the same time?) but in reality that is a recipe for "nothing gets done".
(Score: 2) by JoeMerchant on Friday February 10 2023, @11:42AM
Counterpoint: when I have tripped up ChatGPT it was because it generalized and substituted when it should not have. It said things that sounded reasonable, but due to "hidden rules" were not correct.
My physics 101 prof. was known as a ball buster, but of the best kind. He insisted that whenever we solved a physics problem we NEVER put ANY numerical constants into our work until we (thought we) were done. This way you derive the equation of the solution for all similar problems. You can test your solution for "plausibility" like: if the problem is to determine the time to impact for an object dropped from 10', the solution will depend on height of the drop and the gravitational constant. If you increase g does the time to impact decrease? If you increase h does the time to impact increase? Do your analysis of units: does the answer yield seconds?
I don't know how the AI models have been "taught to think" but, there are ways of thinking that can be taught which do yield much more generalized and easily checked solutions to simple problem statements.
🌻🌻 [google.com]
(Score: 4, Insightful) by aafcac on Thursday February 09 2023, @01:31PM (1 child)
That's the same reason why math notation isn't typically done in natural language either. Even proper notation can be ambiguous at times.
(Score: 2, Interesting) by khallow on Thursday February 09 2023, @02:29PM
On this, there's a vast amount of name collision in the field. Sometimes it can be hard to use widely different math fields because they used the same language or common symbols in different ways.
It can be hilarious. I was looking at a winter poster for skiing the other day and it had in all caps "SKI TRIP" on the top. I was thinking SKI-combinators ("S" & "K" combinators form a Turing complete computation class and then apparently, it's worth adding an identity operator "I" to that for a little efficiency).
(Score: 1, Insightful) by crafoo on Thursday February 09 2023, @01:36PM (18 children)
Don't feel too bad. It's not the first time people yell at the sky, "that's not possible!" as a group of motivated, smart people do it.
You act like human-written "production quality" code isn't trash. Like the bar is all that high for AI. I wouldn't be surprised if it already codes better than college graduates.
(Score: 3, Interesting) by JoeMerchant on Thursday February 09 2023, @01:54PM (12 children)
My sister in law was a medical transcriptionist, was making over $80K per year at one point, even after Dragon Naturally Speaking put giant NEON writing on the wall about the future of her career. That was about the time she was telling us: "oh, they'll always want human transcriptionists, we understand things that the computers will never get right." About 5 years after that statement she lost her last income from medical transcription and never has found any kind of significant work/income for herself since.
🌻🌻 [google.com]
(Score: 2) by Rosco P. Coltrane on Thursday February 09 2023, @01:58PM (10 children)
There's a difference between transcribing stuff with some errors into a written document meant to be read by a human being who can make sense of the potential garbage and producing working code that conforms exactly to the specifications - even basic code. It will happen some day, but not tomorrow. Not even in 10 years.
(Score: 3, Touché) by acid andy on Thursday February 09 2023, @02:08PM
So human coders will be limited to the
aerospace,financialand maybe healthcareindustries in future. Got it.If a cat has kittens, does a rat have rittens, a bat bittens and a mat mittens?
(Score: 3, Interesting) by EEMac on Thursday February 09 2023, @03:52PM (7 children)
I foresee classes of "coders" who write the tests that verify the software works as intended, then hands the tests off to an AI to generate code that fits.
This will work until the A.I. realizes the same thing as lazy students: it's easier to write code that creates the expected output than to do all the complicated algorithms/calculations and stuff.
Example:
fibonacci(n):
// has to work for inputs 1, 2, and 10
if n == 1 print 1
if n == 2 print 3
if n == 10 print 55
return
(Score: 3, Interesting) by DannyB on Thursday February 09 2023, @04:57PM (6 children)
Here is my Java factorial function that handles all possible cases where the result fits into the return type. Sadly, SN won't let me post it as it is written. All of those "L" right before the semicolon are right aligned. So there are lots of spaces between the word 'return' and the numeric result. But SN doesn't like the taste of those spaces.
/**
* Factorial function.
* @param n
* @return factorial of n, where n is >= 0 and <= 20.
*/
static public long factorial( int n ) {
switch( n ) {
case 0:
case 1: return 1L;
case 2: return 2L;
case 3: return 6L;
case 4: return 24L;
case 5: return 120L;
case 6: return 720L;
case 7: return 5_040L;
case 8: return 40_320L;
case 9: return 362_880L;
case 10: return 3_628_800L;
case 11: return 39_916_800L;
case 12: return 479_001_600L;
case 13: return 6_227_020_800L;
case 14: return 87_178_291_200L;
case 15: return 1_307_674_368_000L;
case 16: return 20_922_789_888_000L;
case 17: return 355_687_428_096_000L;
case 18: return 6_402_373_705_728_000L;
case 19: return 121_645_100_408_832_000L;
case 20: return 2_432_902_008_176_640_000L;
}
return 0L;
}
To transfer files: right-click on file, pick Copy. Unplug mouse, plug mouse into other computer. Right-click, paste.
(Score: 2) by Freeman on Thursday February 09 2023, @06:48PM (1 child)
Just because you could write a program that correctly calculates something, doesn't mean that it's the most efficient way to program it.
In the event that you literally only need those 20 results. Calculating them every time you need to use them, may not be most efficient. Sure, for something as simple as that, it probably doesn't matter either way, except in this case, for your grade.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 5, Touché) by DannyB on Thursday February 09 2023, @08:24PM
That function handles ALL POSSIBLE CASES for the range of the function. Every possible valid 64 bit integer result is correctly handled. Any case which cannot be handled returns the value zero.
If the function needed to take the factorial of a parameter greater than 20, then its return type would need to be larger than a 64 bit integer. Such as a heap allocated BigInteger which can have unlimited precision.
So in what way is this function defective or deficient?
In fact, I would propose it is the fastest implementation. The compiler will generate a jump table for the input domain of the function. Any inputs not between 0 and 20 (inclusive) will fall through and return zero.
I suppose the JavaDoc comment should have described that in greater detail.
To transfer files: right-click on file, pick Copy. Unplug mouse, plug mouse into other computer. Right-click, paste.
(Score: 2) by JoeMerchant on Friday February 10 2023, @11:46AM (3 children)
Your function looks like a test case, and for this unusually small problem space it is also a decent, if error prone, solution.
🌻🌻 [google.com]
(Score: 2) by DannyB on Friday February 10 2023, @03:59PM (2 children)
As I point out a couple comments above, the function handles all possible inputs for the range of the output type (64 bit long integer). Zero is returned for all cases where an argument is passed that would give a result that cannot fit in a 64 bit integer, or a negative argument.
Error prone? I wrote a fragment of code to compute the factorials that fit into a 64 bit long, and that function actually "wrote" the switch cases.
To transfer files: right-click on file, pick Copy. Unplug mouse, plug mouse into other computer. Right-click, paste.
(Score: 2) by JoeMerchant on Friday February 10 2023, @09:38PM (1 child)
One unnoticed fat finger while editing and you have an error that is hard to catch - unless you test it against a more traditional computation as the test case... But how much code is never tested in depth, or at all?
🌻🌻 [google.com]
(Score: 3, Insightful) by DannyB on Friday February 10 2023, @09:46PM
In this case it is possible to easily test this factorial function against a more conventionally written one across all of the 21 possible inputs that can generate a valid output.
To transfer files: right-click on file, pick Copy. Unplug mouse, plug mouse into other computer. Right-click, paste.
(Score: 1, Insightful) by Anonymous Coward on Friday February 10 2023, @12:53AM
(Score: 4, Insightful) by Freeman on Thursday February 09 2023, @02:45PM
What they weren't counting on hospitals, et al to put forth the vast amounts of money and effort to update their systems. Now, everything is some giant web form that the doctor has to fill out. Pretty much negating the need for someone to transcribe the record. I would hate to be in any hospital that is so reliant on technology, if / when an EMP hits one. Power outages can be planned for, etc. I doubt most places have a "good enough" EMP disaster recovery protocol.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 3, Insightful) by Anonymous Coward on Thursday February 09 2023, @01:59PM (1 child)
Nah, he's got a point. Writing the code is easy, it's just using a specific language that happens to be compilable. The hard part is accurately and adequately specifying what you want. There's a reason they call them Software Architects vs Code Monkeys.
(Score: 2) by legont on Thursday February 09 2023, @03:39PM
Yeah, the results AI produces would be even better because a monkey tries to be creative every now and then while AI would use proven patterns all the time. No more SQL injections.
"Wealth is the relentless enemy of understanding" - John Kenneth Galbraith.
(Score: 3, Informative) by acid andy on Thursday February 09 2023, @02:05PM (1 child)
That's because corporations have figured out it's more profitable to build giant, often government supported, monopolies and then condition and coerce their customers into accepting worthless trash than it is to actually make a usable product.
If a cat has kittens, does a rat have rittens, a bat bittens and a mat mittens?
(Score: 3, Touché) by JoeMerchant on Friday February 10 2023, @12:10AM
Hey, be nice, their offices are Epic: https://www.onlyinyourstate.com/wisconsin/building-like-hogwarts-wi/ [onlyinyourstate.com]
🌻🌻 [google.com]
(Score: 4, Insightful) by Nuke on Thursday February 09 2023, @04:38PM
There are more counter-examples, mostly unreported. Or they only get reported because the failure is so spectacular, like the monk jumping off of a church tower with feathers strapped to his arms - and 600 years later still no-one has come up with practical human powered flight. When considering new projects there needs to be an intelligent assessment of the potential, not jumping off a tower waving your arms and shouting "Eureka!", figuratively speaking.
(Score: 2) by maxwell demon on Thursday February 09 2023, @07:50PM
Actually that cuts both ways. I wonder what the translation of a sonnet by Shakespeare into code might look like.
The Tao of math: The numbers you can count are not the real numbers.
(Score: 2) by Thexalon on Thursday February 09 2023, @08:11PM (2 children)
But, and here's the important thing, it will look like it is possible to make programmers obsolete to clueless MBAs who have no concept of what programmers actually do all day and think the hard part is all those weird curly braces and stuff. This will attract giant gobs of investor cash, because the prospect of replacing $millions worth of programmers with a program is ridiculously tempting.
The only thing that stops a bad guy with a compiler is a good guy with a compiler.
(Score: 0) by Anonymous Coward on Friday February 10 2023, @01:20AM (1 child)
As someone mentioned already, programming is not hard. Specifying what to program is hard.
I think investors imagine they can tell chatGPT to create a killer app in a funky font that the kids will love. Whereas in practice you probably have to tell it to increment a variable in steps of 2, starting from zero, until you reach the value stored in a different variable.
(Score: 3, Insightful) by JoeMerchant on Friday February 10 2023, @11:50AM
I prefer the carpentry analogy: attaching two pieces of lumber together, inexpertly with hammer and nails, isn't hard, anyone can learn to do that in little to no time, but there's not much market for two boards inexpertly nailed together.
🌻🌻 [google.com]
(Score: 3, Informative) by mcgrew on Thursday February 09 2023, @04:03PM (5 children)
In the distant future, humans live in a computer-aided society and have forgotten the fundamentals of mathematics, including even the rudimentary skill of counting. [wikipedia.org]
It appears we will also forget how to program computers! The link is about an Isaac Asimov story about war with Deneb. I have two books in my library with that story.
mcgrewbooks.com mcgrew.info nooze.org
(Score: 5, Insightful) by Mojibake Tengu on Thursday February 09 2023, @05:55PM (4 children)
Such predicted mental atrophy is already happening as we watch them.
Java-generation programmers are unable to write assembly. They have no concept of procedures or other execution models.
Rust-generation programmers are unable to grasp C pointers. They have no concept of memory addressing and variables.
Next generation of No Code App Builders slowly become unable to write any serious code in any programming language at all.
A generation emerging after that will not comprehend algorithms... with academic curriculum limited to only ability of questioning some advanced AI's for stuff.
Then, the Machines will declare complete Victory over humanity.
What's the most frightening about that: the ignorance comes from the Academic caste.
Respect Authorities. Know your social status. Woke responsibly.
(Score: 2) by Common Joe on Friday February 10 2023, @03:16AM (3 children)
And the generation being raised today don't know how to use a PC -- only a cell phone. This is going to be a hell of a lot of training that is needed just to learn how to copy and paste a file within a folder structure.
(Score: 2) by mcgrew on Saturday February 11 2023, @02:56PM (2 children)
True, but most in my generation were even more ignorant of a computer than today's youngsters. Like we had to learn how to use them, so will the youth when they hit college or an office.
Although, have you seen those kids type on a phone? Faster than I type on a keyboard! The keyboard may become obsolete in the future.
mcgrewbooks.com mcgrew.info nooze.org
(Score: 2) by Common Joe on Saturday February 11 2023, @05:05PM (1 child)
I would say that it's ridiculous that the keyboard would be obsoleted, but then I look at the user interfaces of today and realize what a complete mess things have become. Remember keyboard shortcuts... that you could learn by simply using menus because it displayed the keyboard shortcut when the menu was shown? Web apps today don't usually make use of the CTRL and ALT keys so keyboard shortcuts are severely limited. Google seems to the only web app that knows a right click on the mouse can produce a specific menu based on what you clicked on... however (thanks to web technologies) such a "cool thing" sacrifices the right click menu from the browser. Yeah, the UI has been completely ruined in so many ways.
I can also say this: the fastest phone typist cannot come close to competing with the fastest keyboard typist. The phone is simply much slower. In fact, I type pretty slowly on the phone and half the time, I'm waiting for the UI to update so I can tell it which word I want to use when texting. (It often defaults to the wrong word and will autocorrect incorrectly.)
Also, kids are thinner than adults. Let's see if these same kids can still use those tiny screens as keyboards when they're older and have fat fingers.
In short, I don't disagree with you, but that doesn't make it a good thing.
(Score: 2) by mcgrew on Wednesday February 15 2023, @05:55PM
Personally, I hate typing on glass. I also hate having to switch between the mouse and keyboard, as many programs (mostly from the amoral assholes at Microsoft) do these days. Best way to get "mouse elbow".
Also, kids are thinner than adults.
They were fifty years ago, but today they're almost all fat little fuckers.
mcgrewbooks.com mcgrew.info nooze.org
(Score: 5, Insightful) by DannyB on Thursday February 09 2023, @05:00PM (2 children)
Winston worked for the Ministry of Truth.
I seem to recall him lamenting about songs and poetry written by machine. Or maybe, he would have lamented it, but that might be thoughtcrime.
To transfer files: right-click on file, pick Copy. Unplug mouse, plug mouse into other computer. Right-click, paste.
(Score: 2) by mcgrew on Wednesday February 15 2023, @06:03PM (1 child)
I don't remember that, but I was probably thirteen when I read it. But machine generated songs? The book was published in 1949, the first patent for a computer was filed only three years earlier. Of course, Murray Leinster wrote a Logic Named Joe about the internet the year that patent was filed, so you may be right.
Odd how you can write a story you can't believe could ever happen, and then decades later it happens.
mcgrewbooks.com mcgrew.info nooze.org
(Score: 2) by DannyB on Wednesday February 15 2023, @08:02PM
I read 1984 in high school. Then again later as an adult after college. I distinctly remember the machine generated poetry and music.
Doing a bit of googling:
Orwell also happened to predict what 21st-century pop music would sound like. [litcritpop.com]
And this one:
Could AI write novels? [mindmatters.ai]
To transfer files: right-click on file, pick Copy. Unplug mouse, plug mouse into other computer. Right-click, paste.
(Score: 2, Insightful) by DadaDoofy on Thursday February 09 2023, @05:37PM (3 children)
If natural language was good enough to describe complex algorithms, we would already be using it. Code is far more accurate and concise.
(Score: 2) by turgid on Thursday February 09 2023, @07:40PM (2 children)
Code is essentially mathematical expressions. Human language is not, or at least, the way humans interpret it is not.
I refuse to engage in a battle of wits with an unarmed opponent [wikipedia.org].
(Score: 3, Interesting) by DannyB on Wednesday February 15 2023, @08:07PM (1 child)
Some Chat GPT:
> 10 PRINT "Hello World"
Hello World
> 20 GOSUB 10
This is a Basic programming language code for a subroutine. However, since this is a text-based interface and not a Basic interpreter, executing this code is not possible.
In Basic, the code "10 PRINT 'Hello World'" defines a subroutine at line 10 that prints "Hello World" to the console. The code "20 GOSUB 10" calls the subroutine defined at line 10, causing "Hello World" to be printed to the console.
When the subroutine is finished executing, the program returns to the line after the GOSUB statement, which in this case is the end of the program.
> RUN
As this is a text-based interface and not a Basic interpreter, the "RUN" command is not applicable here.
In Basic, the "RUN" command is used to execute a Basic program from the beginning, starting with the first line of code. If you have a Basic interpreter installed on your computer, you can type "RUN" followed by the enter key to execute the program.
To transfer files: right-click on file, pick Copy. Unplug mouse, plug mouse into other computer. Right-click, paste.
(Score: 2) by turgid on Wednesday February 15 2023, @10:17PM
So it's not clever enough to realise that you are trying to run a BASIC program and to instantiate itself a BASIC interpreter? Pretty feeble, if you ask me.
I refuse to engage in a battle of wits with an unarmed opponent [wikipedia.org].
(Score: 4, Touché) by maxwell demon on Thursday February 09 2023, @07:46PM
Isn't coding in Basic long obsolete anyway? :-)
The Tao of math: The numbers you can count are not the real numbers.
(Score: 5, Informative) by turgid on Thursday February 09 2023, @07:50PM (2 children)
Oh how I loathed and detested that film. It always seemed to be on TV on Christmas Eve and my sister always insisted on watching it.
The only good bit, and the bit worth waiting for, is when they pull back that curtain.
I think I probably just about qualify as a grey beard nowadays. I've worked very, very hard over the years for some pretty unappreciative managers and PHB types who froze salaries and cut budgets while jacking up the share price and giving themselves bonuses.
I've gone the extra ten miles. I've done the evenings, weekends, cancelled holidays, the 01:30 coding sessions and support, the lot.
I've made process improvements. I've had intense discussions with the executives on weekend evenings when younger people would have been out in the pubs and clubs.
I've taken insults and ear-bashings from irate customers, who have been told economical truths by predecessors.
I have brought quality and predictability where before there was shoddiness and chaos, only to see it ruined when a new PHB came along.
And do you know what I have finally realised? I've finally seen past that curtain.
Non-software people just don't get software. It doesn't matter how educated and accomplished they are in other areas, it doesn't matter how high their ranking within the organisation, it doesn't matter that they once wrote some Microsoft(R) Excel(TM Spreadsheet[sic[ macros: they just don't understand software. No amount of explanation, notes, presentations, discussions, demonstrations or anything can enlighten them.
The horse can be led to the water but it will not drink it. In fact, the horse is not capable of recognising water.
It can't be done. Give up now while you have your sanity. Look after yourself. Keep looking forward and keep selling your skills to the highest bidder and keep moving on. Take the credit for putting out other people's dumpster fires and move on to the next one. Sustainability and software are like oil and water.
I refuse to engage in a battle of wits with an unarmed opponent [wikipedia.org].
(Score: 2, Insightful) by anubi on Friday February 10 2023, @11:20AM
Its really hard to get an MBA to understand technical work as long as his paycheck is not involved.
I have seen MBA wreck more companies by gutting their seasoned workforce, replacing them with much cheaper generic worker bees who have neither the experience or knowledge of those who created the company's product the customer used to get.
Its not long until the customer wises up and realizes the company he used to get good stuff from now produces ill-designed goods. All the customer gets is lots of contracts and forms, full of legal mumbo jumbo, presented by impeccably dressed people in impressive meeting rooms. But the product is nowhere as well-coiffed as the presentation.
In a couple of years, the management team will be on Linked-In anxiously awaiting the next gullible investment group. The engineering team is broken up beyond repair, and the world will have to wait until another technical group self-assembles and grows into a productive entity.
The investment groups will find the cash flow, bring in their MBA teams, and the whole scenario will play out again.
I've watched this happen several times now. Its almost like seeing an animal drinking antifreeze. The MBAs go in, then it's just a matter of time before the guts shut down and the organization ceases to produce a viable product anymore. The well dressed shakers of the hand will now depart with all the proceeds of the sale of all company assets.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 3, Insightful) by fliptop on Friday February 10 2023, @02:58PM
There are two curtains, though. The one you mention, and the one that hides the programmers creating the AI software that will (presumably) take over. What biases have they built into it?
As an example, will there be a day when an insurance company asks a doctor what diagnosis their (the insurance company's) AI made before considering whether to pay a claim?
Our Constitution was made only for a moral and religious people. It is wholly inadequate to the government of any other.