Automating programming: AI is transforming the coding of computer programs

https://www.economist.com/science-and-technology/2021/07/07/ai-is-transforming-the-coding-of-computer-programs

by edward • 5 years ago

118 173 5 years ago

ollien3• 5 years ago

I can't read past the fold here (paywall), but assuming this is about Copilot, it feels a bit disingenuous to suggest it's already "transforming" software development. It's a cool tool, but it's not even available to everyone.

sitkack• 5 years ago

Judging from the turmoil both here and on twitter about Copilot, it has already made a huge impact.

Two more papers down the line.

ollien3• 5 years ago

Impact? Sure. Especially given the controversy around how it's trained. I just don't know that I'd suggest it's changed software development as we know it.

cjg• 5 years ago

https://archive.is/bTkFX

ollien3• 5 years ago

Thanks for the link! The article feels a lot closer than the headline/first two paragraphs to what I actually see going on.

ok2938• 5 years ago

One thing to keep in mind: With every request-response cycle for the probably upcoming paid service, you are feeding constant data points to a commercial entity. Basically your typing habit, the project development over time, the bugs, the fixes, all will be recorded.

Think about it, one well paid profession in the grips of capital siphoning off valuable, highly intellectual work (by using "open source" already), repackaging it and the selling it back to the people it aims to replace.

Imagine a million lawyers' days being recorded in 60s intervals - including what they type, the files they have opened, etc. - how long would it take to encode the average lawyer into a hdf5 file? How many lawyers would be fine with that kind of assistance?

If that's not ingenious, then I do not know what is.

Call to action: Programmer, wake up.

Edit: Just to address some typical rebuttals - I'm all for progress and cool toys, but I'm all against monopolies and mega-corps that control and monitor every aspect of your life.

i_haz_rabies2• 5 years ago

Programmers have the worst intelligence to gullibility ratios of any profession.

* I should elaborate. We tend to easily accept the "this work is meaningful" line, we volunteer our time because "programming is a craft," and we will hop on tooling bandwagons like this because it's shiny and new without considering the ramifications.

_y5hn2• 5 years ago

This will sell like hot cookies. The only thing that can destroy it will be the immense cargo-culting. So software developers are safe, for now.

staticassertion3• 5 years ago

how long would it take to encode the average lawyer into a hdf5 file

I'd say, minimally, a good 50 years, except perhaps for the busywork. At that point you could maybe have a tool that could handle a lawyer's job. Law's often an extremely creative, verbal, expressive job, so I think machines are particularly unlikely to do well there.

I'm not worried about being automated away. If someone wants to make coding easier, neat, I'm not a programmer, that's just how I get shit done. I solve complex problems, and I don't see machines taking that job for a long, long time.

Frankly, if machines can take away the mechanical bullshit part of a job and let us humans do the interesting, creative work, that works for me.

kwhitefoot3• 5 years ago

Frankly, if machines can take away the mechanical bullshit part of a job and let us humans do the interesting, creative work, that works for me.

How much "interesting, creative work" is there? Surely most of the work that most of us do, and that most of us are capable of and interested in, is not simultaneously interesting and creative.

I count myself as lucky and I have certainly had a more interesting working life than average but the proportion of that forty years of work that counted as both interesting and creative is probably, looking back on it, measured in single digit percentage points.

If a machine does all the easy stuff how would I ever get enough practice to stretch myself to the difficult creative parts?

swiley• 5 years ago

It's not surprising that "normal users" don't think about this but what I've found extremely surprising is that otherwise fairly skilled programmers don't even think about it.

auggierose• 5 years ago

The more skilled the less there is to think about.

bluetwo• 5 years ago

A good lawyer, like a good programmer isn't valued just on their output, but on the questions they ask that lead them to the correct output. This is much more difficult to encode.

Jeff_Brown9• 5 years ago

It's an interesting comparison, law and code. GitHub's AI gets a sense of what you're trying to do, and tries to move the ball forward. Sometimes the context it has is enough to do that. It's better with verbose languages.

Law can be pretty verbose. Given half a paragraph, how often can an intelligence (artificial or natural) deduce what's next?

Of course the coding AI has the advantage that coders write (and publish) comments.

booleandilemma• 5 years ago

Programmers are a weird bunch.

Look at stackoverflow for example. Thousands of programmers giving away their professional advice, spending their valuable time…for what exactly? Imaginary internet points?

Open source is another one: here’s my code that I’ve worked so hard on, spent countless hours on - please have it for free, my labor is worthless.

It’s like they’re actively seeking to bring their own value down.

Jeff_Brown9• 5 years ago

The only explanations I can think of are (1) altruism, (2) intent to profit from one's reputation, and (3) simply enjoying puzzle solving. Am I missing any? And are any of them so strange?

sidlls3• 5 years ago

And then they complain bitterly when commercial entities use it without contributing back, treat workers poorly, and all that. It's not weirdness; it's ignorance and/or stupidity.

jhgb• 5 years ago

please have it for free, my labor is worthless.

That's weird logic. Presumably you wouldn't create something you consider worthless - you'd create it because you needed it.

sudosysgen2• 5 years ago

Programmers aren't unique in thagt. If you create the right conditions, driven people will just work for the sake of it, if it can help someone or create something beautiful. It's the reason why us humans are where we are to begin with.

imnotlost• 5 years ago

Philip Glass talks about this exact thing here:

https://thecreativeindependent.com/people/philip-glass-on-co...

meh99• 5 years ago

Humans were giving away their service to the church for promises of forever life.

Mass delusions are easy to propagate as they require little emotional nuance; a salute, tattoo, incantation… those used to be enough to bind large groups.

Groups are out though. We’re leaning into enabling humans as atomic agents and automating away the utilitarian social obligations, and work of biological survival to get there.

jasode• 5 years ago

Thousands of programmers giving away their professional advice, spending their valuable time…for what exactly? Imaginary internet points?

Before Stackoverflow, people were answering questions on USENET newsgroups without any award of points.

And before USENET and the internet, hobbyists would gather to trade advice on building home computers. E.g.: https://en.wikipedia.org/wiki/Homebrew_Computer_Club

The common theme isn't the points... it's that people like to be helpful and share knowledge.

Open source is another one: here’s my code that I’ve worked so hard on, spent countless hours on - please have it for free, my labor is worthless.

There are multiple motivations for open source. In my case, I did it because I could get more value from the community's enhancements than the "free code" the community got from me.

It’s like they’re actively seeking to bring their own value down.

Neither of your 2 examples look like erosion of labor value. On the contrary, it shows that not every activity associated with programming has to be driven by the exchange of money.

mftb• 5 years ago

"Man cannot stand a meaningless life", Carl Jung. I've spent twenty-five years coding, solved some problems, made some money for myself, more for others, but what have I accomplished? It's unlikely it will be through a stackoverflow answer or a post on HN, but I quietly maintain the hope that maybe someday, through Free Software or otherwise sharing my knowledge and code, I might actually accomplish something meaningful. The economic value you're talking about has little meaning for me.

dave_sid2• 5 years ago

So so true. Imagine a plumber turning up at your house and fitting you a new bathroom as their open source project. But then again, if they charged a extortionate ‘support’ fee for years thereafter for doing nothing then that wouldn’t be as crazy. A bit like Red Hat in that sense.

otabdeveloper42• 5 years ago

Programming isn't labor.

Or, at least, it's only "labor" in the same sense that literature or painting is labor.

Does it seem strange to you that writers or painters would want the world to see their work, even for free? It doesn't seem strange to me.

Programming is the same.

smegma2• 5 years ago

Many things are not zero-sum, for example if I help you and you help me we will both be better off than before. Sites like stackoverflow make programmers more productive in aggregate; we would be worse off without them.

If you can't understand why someone would decide to offer their help for free then at least try to be thankful.

icoder4• 5 years ago

Well there's many reasons. But in addition, most on Stack Overflow is the boring stuff: getting a certain operation, task, calculation right, often in a particular language, with a particular framework. The actual development, in my opinion, starts after that.

Just take my day today: interpreting the client's specs and internally communicating and planning a minor change in the context of a few parallel release paths, discussing a few tweaks to a piece of code with a colleague carefully weighing effort, impact on partially rolled out tests, and future flexibility (making best guesses as to where things are going), and analysing / fixing an issue in a piece of code which design delicately balances simplicity, robustness and performance.

Call me naive but I don't see an AI taking over before reaching (or nearing) general AI. Nor do I see how someone that's not a competitor already become one just by having access to Stack Overflow, or even a tenfold of it.

X6S1x6Okd1st• 5 years ago

There are other frameworks than strict personal net worth maximization.

MeinBlutIstBlau• 5 years ago

The heart of the open source movement had more to do with companies locking down software with operating systems and less about star-trek socialism for code. RMS wasn't just some hippie, he actually stuck to his guns and really "stuck it to the man" in a way that actually revolutionized programming today.

ascotan• 5 years ago

Going to drop this Dan Pink video here: https://www.youtube.com/watch?v=KgGhSOAtAyQ

Basically what motivates good engineers is a good problem. If we can meet our basic needs we give the rest away for free. Companies/entities have been using/exploiting this for years. The gamification points are lame (I agree) but I believe there is a general desire to contribute to the common good and give back to the community.

robotresearcher• 5 years ago

my labor is worthless

The opposite is true. Analogy: Volvo invented the three point seatbelt. To maximize the value of the invention, they gave it away free to use.

NorwegianDude• 5 years ago

A lot of people uses stackoverflow for the gamification. Others just want to give something back to the community.

Open source is a great way to market yourself. If I have some really valuable code, then I'll keep it for myself. If I think it would be more useful and gain traction as an open source project then I might publish it and in that way position myself as the top expert on that software.

A lof of businesses are actively using, publishing and contributing to open source. Not because they don't want to make money, but because it's the way to make money.

michaelmrose2• 5 years ago

The betterment of the human species is more important by far than maximizing the profits any particular individual can earn. Once a work is created the marginal cost of reproduction is close to zero. Software is valuable. A copy of software is paradoxically worth nothing.

It's true that software developers need to eat too but we need to figure out a way to do so without making the world more cumbersome, transactional, and shitty.

b3morales• 5 years ago

Both Stack Overflow and publishing your code openly are strongly motivated by pride/showing off your know-how.

When I participated on Stack Overflow, for me it was a combination of learning a lot by answering other people's questions and getting a kick out of demonstrating my expertise.

mr_toad2• 5 years ago

Information wants to be free.

kp995• 5 years ago

Call to action: Programmer, wake up.

The article also mentions - "head of AI products at the St Petersburg office of JetBrains, a Czech developer of programming software, sees time savings of 10% to 20%. "

and the subheading says - "The software engineers of the future will, themselves, be software"

um, how seriously should we consider the fact that "skilled" software engineer/programmer solving really complex problems are at risk?

Can't say if making just another simple CRUD app is not at risk but the statements and their propositions made in the article are too generalize too plant a seed of insecurity in oneself.

icoder4• 5 years ago

Call me naive but I really don't see an AI take over my (senior) software development work, there's just so much to it, including a lot of communication and creativity.

I dare posing that by the time AI takes over my job, it's gotten so advanced that I've got much more to worry about than just my job.

8note• 5 years ago

Its not going to find out the actual work I do, which is largely offline

Once I've solved a problem, I document it in code. Only seeing the code at the end misses the problem solving parts

na85• 5 years ago

Programming has always been destined to become more and more "blue collar", for lack of a better term, pretty much from the outset.

mr_toad2• 5 years ago

Greg Egan wrote a short story about a financial salesperson who’s job was taken by an AI that was trained by listening to his phone calls with clients.

As a programmer though, I’m not worried. Every line of code is generating more code. Every recursive function, every loop, every piece of meta programming. Every script file that calls a Python program that calls c routines that microcode that call actual ops. It’s already all generated code generating more generated code.

When (if) we have full AGI that is capable of truly understanding a real world problem and kicking off a chain of code generation and is affordable, then I’ll be worried.

908B64B197• 5 years ago

Imagine a million lawyers' days being recorded in 60s intervals - including what they type, the files they have opened, etc. - how long would it take to encode the average lawyer into a hdf5 file? How many lawyers would be fine with that kind of assistance?

Pretty sure that large law firms already automate a lot of the work they do. Standard contracts they customize for instance.

But really, aren't a lot of (billables) hours spent explaining legal concepts to customers?

loloquwowndueo• 5 years ago

https://outline.com/bFWAyk

It is about copilot (which it sneakily doesn’t mention by name) and tabnine and also generally about gpt-3 used in the realm of coding. A lot of layperson-oriented phrasing around: “ Microsoft, an American software giant, released a new version of an AI-completion feature which it embeds in coding software called Visual Studio.”

js84• 5 years ago

I think Copilot is as helpful to programmers as ELIZA is helpful to psychiatrists.

I don't think it really understands code, only very superficially (unlike GPT-3 which might actually genuinely understand parts of language, computer code requires more abstracted mental model). ELIZA was similar, some people also believed that it can understand what they are thinking. Turns out it was all just in our heads.

qsort• 5 years ago

I'm not even remotely worried for my job security, but I don't think the ELIZA comparison is apt.

ELIZA is clearly a toy, Copilot is something that could become integrated with the workflow of a professional developer. Maybe not copilot specifically, but if Intellisense is in our IDEs already, something like it is not a stretch.

I don't think it really understands code

It clearly doesn't, but (a) nobody actually claimed that outside of the usual media circus, and (b) that's a very high bar. It doesn't need to understand your code in order to be useful.

js84• 5 years ago

ELIZA wasn't "clearly a toy" when it appeared. It actually confused people.

I don't deny that good suggestions are useful in coding. I am just not sure you can get these by mimicking other people's code without really understanding it.

I think the reason why code is different than natural language is that in code, the meaning of identifiers (words) is very dependent on the context of the specific codebase, and not nearly as universal as in natural language. Sure, there are words like personal names in natural language that change meaning depending on wider context, but they are not as common in the natural text.

So focusing on the names only, rather than for example understanding types of runtime objects/values, and how they change, can be actually actively harmful in trying to understand the code and making good suggestions. So I would believe that suggestions based on e.g. type inference would be more useful.

There is another aspect to this, text has inherently low entropy, it's somewhat easy to predict the next symbol. But if code has low entropy (i.e. there are universal patterns that can be applied to any codebase in a particular programming language), I would argue that this is inefficient from the programmer's point of view, because this entropy could probably be abstracted away to a more condensed representation (which could be then intelligently suggested based on semantics rather than syntax).

In other words, I think the approach assumes that the probability of the next word, based on syntax, is similar to probability of the next word, based on semantics, but I don't believe that's the case in programming languages.

ska• 5 years ago

ELIZA is clearly a toy,

That's a red herring. The "Eliza effect" has nothing to do with how toy-like ELIZA actual was, it is a lesson about how humans were very bad at evaluating its actual capability and insight. This continues to be true today, and is an effect with at least some impact on essentially any AI/ML system that people interact with.

feoren2• 5 years ago

Copilot is clearly a toy.

wongarsu• 5 years ago

unlike GPT-3 which might actually genuinely understand parts of language, computer code requires more abstracted mental model

Shouldn't computer code be simpler to understand and express than natural language? The amount of knowledge about the world is much smaller and mostly concerns artificial systems. And of course programming languages are laughably simple and well defined compared to the constantly evolving mess that is the English language.

js84• 5 years ago

In programming languages, it is much more common to define new words (symbols) than in natural language. The identifier (name) of the symbol (word) in such case is less informative about its nature than the actual definition, and I think this is the Achilles heel of the method.

lambdaba3• 5 years ago

What would it mean to "understand" it. Imagine a program written by such a system, it's easy to imagine an immensely complex question about the system that sounds simple, for instance asking about design choices, code architecture, aesthetic choices like expressivity. Even if it could produce some code, we couldn't say it understood it if it couldn't answer those questions, and those are very difficult questions which would require human-level intelligence, and even more a human psyche.

throwawaygh4• 5 years ago

Shouldn't computer code be simpler to understand and express than natural language?

Yes, but the ideas that are being expressed in formal languages are much more complex and detailed than the ideas that are being expressed in natural language.

Writing some plausibly on-topic blog posts is way easier than writing even a very simple program, especially if you don't really know what you're talking about.

falsaberN1• 5 years ago

I have played with GPT3 and similar lesser models for a while, and in my opinion it doesn't understand anything. It's good at correlation (it might figure out a pikachu is a pokemon or that The Wall is a music album) which is kinda amazing but ultimately seems to be really "useless"* and it's not deterministic.

I think the current "AI" models will never understand a thing unless they become able to update their own models and have some guiding logic to them. Sure it has the context tokens so it might give a little illusion of following along, but the randomness of it all also means it will get wrong stuff that it got right before (and vice versa) and will clearly break the illusion when prodded ("pikachu is a purple dragon", "The Wall was composed by Metallica"). It's unfortunate.

GPT3 and other text models have no sense of state. You can have the same instance work with 2 (or more) different persons and text provided by user A will have no bearing on anything user B does, since it's just trying to complete the provided text in form of tokens. As computationally intensive as it is, the whole thing is a lot simpler than it seems.

It's like we managed to get the "learning" part of intelligence more or less right, but nothing else yet. I always wanted to have an AI "trusted assistant" kinda thing, but even the biggest GPT3 models don't give me much in terms of trust. Wouldn't trust it to turn a lamp on and off as it is. (I know GPT3 is just a text model but you know what I mean). And if it happens, it won't be run locally because the hardware requisites for this stuff are obscene on a good day, which opens a few cans of worms (some of which have been opened already).

*useless as in getting anywhere close to intelligence. It can become decent entertainment and might have some interesting uses to correlate datum A to datum B/C/D, recognition and so, but so far it's just fancy ELIZA. An "educated guess engine" at best.

pydry3• 5 years ago

The software engineers of tomorrow will themselves be software

Is it me or is the Economist much more unabashed these days about representing the fevered dreams of business elites? (as opposed to engineers who know wtf theyre talking about)

This copilot circus reminds me of a restaurant I used to go to went to shit in ~2015 because the owner thought giving iPads to customers meant they could cut payroll by 20%.

Which was very much in vogue at the time:

http://images.gawker.com/itqtvwbe3c0skb99wirm/c_scale,fl_pro...

gruez• 5 years ago

This copilot circus reminds me of when a restaurant I used to go to went to shit in ~2015 because the owner thought giving iPads to customers meant they could cut payroll by 20%.

What went wrong? At some fast food places the ordering machines definitely displaced some of the cashiers. I'm surprised customers wouldn't go for tablets if it meant they didn't have to pay the 15% tip.

pydry3• 5 years ago

Orders took forever or got lost, etc. The servers looked stressed as hell and were running everywhere.

The prices werent any lower they just had fewer staff. Presumably the owner thought they could collect higher profits.

rootusrootus2• 5 years ago

At least in my experience, what went wrong is that ordering machines didn't really improve the experience. Maybe compared to a bad cashier they're preferable. But other than that, they are much slower and less convenient than just telling another human what you want.

To use a fast food example, I somewhat regularly go to a McDonalds when we road trip (I have kids, don't shame me) and they have ordering screens. It takes easily 5x as long to answer all of it's questions about permutations on the meal I want than to just say "gimme a #2 with Dr Pepper, no pickles" to a cashier.

Maybe if they put in a proper voice assistant it would be better. But the touchscreen is not in any way a superior experience.

goatlover• 5 years ago

Someone still has to bring the food out, handle requests not on the tablet along with complaints, and bus the tables. Tip doesn't apply to fast food, so that's not a good comparison.

dragonwriter• 5 years ago

I'm surprised customers wouldn't go for tablets if it meant they didn't have to pay the 15% tip.

They don’t have to tip to start with. And the ones with the most problem with it won't, or will tip less.

And people go to places that aren't fast food because, at least in part, of service, which iPad’s don't provide the way servers do.

rootusrootus2• 5 years ago

I'm surprised customers wouldn't go for tablets if it meant they didn't have to pay the 15% tip.

It seems to have become quite normalized to put tip jars out for all sorts of businesses that aren't really service-based. Setting the expectation. Or putting on a tip line on the credit card receipt -- sure, you don't have to tip, but we want you to think you're abnormal if you don't.

I don't think switching to tablets would do anything to reduce the expectation of tips.

And it hasn't been 15% for years.

cat199• 5 years ago

yes, now with copilot we will be doing visual object oriented programming on our GUI only tablets, and typing code, laptops, and desktops will be a thing of the past - the singularity is in fact here!

Jeff_Brown9• 5 years ago

Indeed. A more accurate title would be "A new AI product might shave 15% off some developers' time spent coding, and some otherr companies hope to cash in too."

pydry3• 5 years ago

Even 15% is way too optimistic.

I'm not even convinced that the effect will be positive. A habit of blindly accepting copilot suggestions could lead to more bugs.

primaryobjects2• 5 years ago

Interesting timing, as tomorrow I will be presenting on this very topic.

"AI programmer: autonomously creating software programs using genetic algorithms"

https://dl.acm.org/doi/10.1145/3449726.3463125

neatze• 5 years ago

So awesome!, rip my weekend plans, I have done similar thing for fun using python and modifying class dictionary (half working), more in scope of multi-objective optimization.

AnimalMuppet2• 5 years ago

Interesting. Would you mind telling me what the largest (in lines of code) working program it has created is?

primaryobjects2• 5 years ago

One of the largest programs generated had consisted of about 300 instructions.

This particular program included if/then conditionals, counting down in a loop, concatenation of a numeric value with text, and displaying output.

The system also attempts to optimize the number of programming instructions executed, in which case complexity can actually be considered based upon the resulting behavior, rather than LOC.

awaythrowact• 5 years ago

Did PyTorch increase or decrease demand for ML developers?

Did Ruby on Rails increase or decrease demand for web developers?

Did C increase or decrease demand for systems programmers?

Automation and abstraction is a complement, not a substitute.

If CoPilot works and we get to focus on higher level goals and developer throughout increases, it will be good for developers and good for the world. I hope CoPilot succeeds.

sly010• 5 years ago

AI has an issue that AI assistants tends towards mediocrity.

In the mid-term this is great for sub-mediocre AI users who seemingly get a boost in performance/quality, etc, but people tend to not recognize that in the long term there is a dangerous feedback loop. AI models just regurgitate their training, pretending it's state of the art.

Every now and then people bring up the fact that relevant long tail Google results are no more. Every time we click through on the first page, we are collectively training google to be mediocre and in exchange it's training us to be the same.

We had the same debate around criminal profiling, insurance, credit ratings, automated censorship, even 'art'.

Programming won't be that different. It will be even useful in a limited sense. AI will give a lot of mediocre results for mediocre questions and the wrong answer for 'new' questions, but we won't know which one is which. But remember that in the long term we are feeding the outcome back to the models, so it's not real progress, it's just a proliferation of "meh".

warkdarrior• 5 years ago

AI has an issue that AI assistants tends towards mediocrity.

Actually humans have an issue here, since the majority of humans are mediocre at what they do (Bell curve and all that). Thus AI assistants threaten to replace the majority of humans.

neonate• 5 years ago

https://archive.is/bTkFX

feoren2• 5 years ago

Copilot appeals to bad developers, which it's going to learn from. It's been trained on a codebase that is mostly bad and/or outdated. It will produce code that is usually bad and the bad programmer using won't properly review and fix it, so it will never learn how bad it is.

If copilot replaces anything, it will be the legion of failed developers that end up working for Oracle and Comcast currently doing god-knows-what with all their time. Those systems are already giant festering mounds of technical debt; imagine how bad it will get when literally no human has ever seen the code producing that weird bug.

hkt2• 5 years ago

In all likelihood, it'll be much the same as when literally only one human has ever seen the code producing that weird bug...

salawat• 5 years ago

My dream in life is to work myself out of a job. To take a bunch of the hard frustrating stuff out of things, the next guy doesn't have to suffer through solving the same problems.

With programming, solutions aren't physically quantized except in the sense sufficiently skilled person hasn't run into an articulation of problem and solution they can actually understand yet.

Programmers aren't an odd bunch. We just like writing encyclopedias that people can read, understand, and build off of so someone can find something cool to do with it.

There's just a point where as system appreciating people, we can see where our own work is obviated or worked against if you just toss it at a machine that mangled it in such a way where the underlying ideas and knowledge is lost.

oliv__• 5 years ago

Well, great maybe the AI can create and run a few more successful SaaS businesses for me

ausbah2• 5 years ago

co-pilot seems great at automatically writing boilerplate code, but iffy at doing anything novel - like most supervised learning systems

YeGoblynQueenne9• 5 years ago

This article of course presents automatic programming as if it's a brand new thing that is only now possible to do thanks to "AI", but that is only how companies that sell the technology find it convenient to present things to market their products. The thing to keep in mind is that program synthesis is an old field and a lot of progress has been made that is completely ignored by the article. Of course most people haven't even heard of "program synthesis" in the first place, simply because it is not as much hyped as GPT-3 and friends.

From my point of view (my field of study is basically program synthesis for logic programs) code generation with large language models is not really comparable to what can be achieved with traditional program synthesis approaches.

The main advance seems to be in the extent to which a natural language (i.e. English) specification can be used, but natural language specifications are inherently limited because of the ambiguity of natural language. Additionally, trying to generate new code by modelling old code has the obvious limitation that no genuinely new code can be generated. If you ask a language model to generate code it doesn't know how to generate - you'll only get back garbage. Because it has no way to discover code it doesn't already know how to write.

So Copilot, for example, looks like it will make a fine boilerplate generator - and I'm less skeptical about its utility than most posters here (maybe partly because I'm not worried my work will be automated) but that's all that should be expected of it.

montenegrohugo• 5 years ago

Hmmm. A few points.

First, I've never heard of program synthesis, and it seems like an interesting topic. Could you point me to some resources so I can learn more about it?

Second. I take issue with this statement:

"trying to generate new code by modelling old code has the obvious limitation that no genuinely new code can be generated"

I disagree. We've seen GAN's generate genuinely new artwork, we've seen music synthesizers do the same. We've also seen GPT-3 and other generative text engines create genuinely interesting and innovative content. AI dungeon comes to mind. Sure, it's in one way or another based on it's training data. But that's what humans do too.

Our level of abstraction is just higher, and we're able to generate more "distinct" music/code/songs based on our own training data. But that may not hold in the long term, and it also doesn't mean that current AI models can do nothing but regurgitate. They can generate new, genuinely interesting content and connections.

throwawaygh4• 5 years ago

First, I've never heard of program synthesis, and it seems like an interesting topic. Could you point me to some resources so I can learn more about it?

I'll leave it to the GP to give a lit review, but will say that CS has (always had) a hype-and-bullshit problem, and that knowing your history is good way to stay sober in this field.

We've also seen GPT-3 and other generative text engines create genuinely interesting and innovative content.

Making things humans find entertaining is easy. Markov chains could generate genuinely interesting and innovative poetry in the 90s. There's a Reply All episode on generating captions for memes or something where the two hosts gawk in amazement at what's basically early 90s tech.

They can generate new, genuinely interesting content and connections.

Have you ever dropped acid? You can make all sorts of fascinating content and connections while hallucinating. Seriously -- much more than when you're sober. Probably shouldn't push to prod while stoned, though.

Art and rhetoric are easy because there's really no such thing as "Wrong". That's why we've been able to "fake creative intelligence" since the 80s or 90s.

Almost all software that people are paid to write today is either 1) incredibly complicated and domain-specific (think scientists/mathematicians/R&D engineers), or else 2) interfaces with the real world via either physical machines or, more commonly, integration into some sort of social processes.

For Type 1, "automatic programming" has a loooong way to go, but you could imagine it working out eventually. Previous iterations on "automatic programming" have had huge impacts. E.g., back in the day, the FORTRAN compiler was called an "automatic programmer". Really.

For Type 2, well, wake me up there's a massive transformer-based model that only generates text like "A tech will be at your house in 3 hours" when a tech is actually on the way. And that's a pretty darn trivial example.

There is a third type of software: boilerplate and repetitive crap. Frankly, for that third type, I think the wix.com model will always beat out the CoPilot model. And it's frankly the very low end of the market anyways.

dreamcompiler• 5 years ago

First, I've never heard of program synthesis, and it seems like an interesting topic. Could you point me to some resources so I can learn more about it?

Look up the work of Cordell Green. He invented the idea and built a company around it.

https://www.kestrel.edu/people/green/

YeGoblynQueenne9• 5 years ago

interesting topic. Could you point me to some resources so I can learn more about it?

The following is a comprehensive introduction and survey of recent work in the field:

https://www.microsoft.com/en-us/research/wp-content/uploads/...

Regarding novelty, neural networks are excellent at interpolation but awful at extrapolation. What I mean by interploation is that the models learned by a neural network represent a dense region of cartesian space circumscribed by the neural net's training instances. The trained model can recognise and generate instances included in this dense region, including instances not included in its training set, by interpolation, but it is almost completely incapable of representing instances outside this dense region by extrapolation [1].

So basically what I'm saying above is that deep learning language model code generators can't extrapolate to programs that are not similar to the programs they've been trained on. Hence, no real novelty. But you can go a long way without such novelty hence my expectation that e.g. Copilot can be a nice boilerplate generator.

______________

[1] My standard reference for this is the following article by François Chollet, maintainer of Keras:

https://blog.keras.io/the-limitations-of-deep-learning.html

See also the second part of the series of articles where possible ways are proposed to address this limitation:

https://blog.keras.io/the-future-of-deep-learning.html

The latter is relevant to the conversation because the author basically proposes augmanting current deep learning approaches with a form of program synthesis.

plutonorm• 5 years ago

Said all of the vision researchers before they got blown out of the water by convolutions neural networks. Adapt or die.

charcircuit• 5 years ago

CNNs can't recognize what they haven't been trained on.

YeGoblynQueenne9• 5 years ago

That's a very low effort, and low quality comment. I'm sorry that I had to read it.

When I post in conversations like this I do it in the hope that my comments will invite robust debate from knowledgeable contributors, that will challenge my assumptions and my knowledge and help me refine my understanding of the subjects I am interested in. This comment is really so much not that, that it hurts.

codetrotter• 5 years ago

This article of course presents automatic programming as if it's a brand new thing that is only now possible to do thanks to "AI", but that is only how companies that sell the technology find it convenient to present things to market their products.

Reminds me of this essay that pg wrote in 2005, titled The Submarine. Someone linked it in another thread here on HN recently.

One of the most surprising things I discovered during my brief business career was the existence of the PR industry, lurking like a huge, quiet submarine beneath the news. Of the stories you read in traditional media that aren't about politics, crimes, or disasters, more than half probably come from PR firms.

http://www.paulgraham.com/submarine.html

perl4ever2• 5 years ago

most people haven't even heard of "program synthesis"

Since my first programming job, I've generated code in whatever language I'm using with that language. SQL from SQL, Perl from Perl, etc. And/or regular expressions.

This page:

https://en.wikipedia.org/wiki/Metaprogramming

leaves the impression that using programs to make programs requires a special language or special features, and it was rather a dead end in AI.

YeGoblynQueenne9• 5 years ago

Metaporgramming is a technique that can be used for program synthesis, but it is not the same as program synthesis. In program synthesis you generate a program from a specification that can be more or less complete, for instance it can be an entire program written in some formal language, or it can be a set of input-output examples of the target program etc. Synthesisers typically search a space of programs (a set of strings in the target language) for a program that satisfies the specification.

I think what you describe is the task of putting together parts of a program according to a configuration. That's a related task and you could certainly see it as a kind of program synthesis, but because it's a common task that we know how to do well, it's not really what program synthesis research is focused on.

Some programming languages are definitely better suited to program synthesis than others. That's particularly the case with languages that have little of what I'd call "ceremony", syntax that's not really necessary to express the behaviour of a program, but are there to help the programmer organise the code, such as classes, maybe, or interfaces, say.

I don't know that metaprogramming was a dead end in AI. I hear that a lot for many subjects of AI research that are still active. I mean, if you asked machine learning researchers ten years ago about neural networks they'd tell you "it's a dead end", or "we tried that and it didn't work" etc. So, unless you're very knowledgeable about a subject of AI research my suggestion is to be cautious when extracting conclusions from a perfunctory perusal of wikipedia or blog posts etc. That's not where most of the knowledge is stored for such deep subjects as AI research.

moyix3• 5 years ago

I think it's inevitable that program synthesis will adopt deep learning methods, whether or not those look like large language models. There's been a bunch of work on this (that you're probably familiar with) from Dawn Song's group:

https://sunblaze-ucb.github.io/program-synthesis/index.html

I think of large LMs trained on code as something that will sample strings from the space of "code humans might write". This isn't nothing, but as you say it's unlikely to produce programs that solve problems no human has solved before. However, assuming you have a rigorous enough spec, I don't see why you couldn't use something like reinforcement learning on such a model in conjunction with traditional verification tools...

Anyway, I agree it's unfortunate that program synthesis has been overlooked in the latest AI code generation craze -- I've mentioned it to a few journalists when talking about things like Copilot but it doesn't tend to end up in the final article.

le-mark• 5 years ago

However, assuming you have a rigorous enough spec,

And thus the ball is moved back to the human programmers court, as it has since Cobols sentences and paragraphs were supposed to allow managers and business analyst to write code.

YeGoblynQueenne9• 5 years ago

I'm aware of the work on neural program synthesis that you link to but it would be unfair to say I'm familiar with it. I've read a few papers they've written, here and there. Interesting work, but as they say on their website, it's an extremely challenging problem.

I'm pretty sure that as you say deep learning will be used for program synthesis, in fact it already has - the article above mentions a few applications and startups etc. I'm really curious to see how all this will pan out a few years from now. But the applications that are developed commercially are really trying to implement programming aids, rather than solutions to the full program synthesis problem, which, in a nutshell, is to be able to generate arbitrary programs from a specification of some sort.

I'm skeptical that deep learning can really solve this more general (and much harder) problem. The reason I'm skeptical is that neural networks are great approximators but they're not known for their great precision, I guess. With program synthesis the point is usually to really get a program that does what the specification says, no ifs and buts and no 80%-there's. I don't know how neural netwokrs can do in this, but I don't think the answer is "very well".

> Anyway, I agree it's unfortunate that program synthesis has been overlooked in the latest AI code generation craze -- I've mentioned it to a few journalists when talking about things like Copilot but it doesn't tend to end up in the final article.

Well, thanks for fighting the good fight :)

You've piqued my curiosity, of course. Can you share one of the articles you mention?

bglazer• 5 years ago

Always happy to see a comment from YeGoblynQueenne on a program synthesis post.

I’m optimistic about the prospects of some combination of logic solvers and deep learning. As you mention, natural language specs are only approachable with a language model. I think logical constraints could function as really powerful regularization on the output of neural network code gen. Like, constraining output to only programs that satisfy logical constraints inferred from the language.

Also, some forms of programming are super non-trivial to describe and are much more easily specified with input output pairs. This is where the more traditional synthesis approaches really shine. Excel flashfill works astoundingly well for this.

YeGoblynQueenne9• 5 years ago

Thanks for the endorsement :)

> Also, some forms of programming are super non-trivial to describe and are much more easily specified with input output pairs.

I agree of course, but then I would, wouldn't I? That's the kind of thing I'm studying...

The Gulwani et al. work behind FlashFill made a huge impression in program synthesis circles when it first came out.

distribot3• 5 years ago

Unrelated but Israel is so cool in terms of technological prowess. A nation of 9 million people and almost every article about cutting edge tech includes them. Really remarkable imo

Animats• 5 years ago

We may get automated programming, but not by the GPT-3 route. GPT-3 is a generator for moderately convincing meaningless blithering. It says something about the nature of discourse that GPT-3 is better than many pundits.

Here's a GPT-3 generated manual for a "flux capacitor".[1] It's a fun read, but meaningless. When the output text needs to integrate a collection of hard facts, GPT-3 is not helpful.

[1] https://archive.is/lQFHC

minimaxir• 5 years ago

In the context of code generation, coding typically follows patterns, which GPT/Transformers is very, very good at identifying.

xaduha• 5 years ago

Wake me up when it starts deleting bad code instead of producing more of it.

distribot3• 5 years ago

I think some of us have our head in the sand about how disruptive this will be to software engineering as an industry. I find the applied stats-yness of ML really boring, but I also want to hedge against some of these companies creating run away hits that make my current job obsolete.

throwawaygh4• 5 years ago

If you're writing the sort of code copilot generates, you should be more worried about things like wix.com, RPA products, WordPress, or high school kids. (No hate, my summer job in high school was PHP CRUD stuff, but it's definitely the burger flipping of the software development world.)

distribot3• 5 years ago

Is there any reason to think there's a ceiling on these types of products or a big distinction between different types of engineering? I understand there will still be software engineers to validate, supervise, etc. But what if every team of 10 people today can be reduced to 3? What will that do to the industry? I have two FANG companies on my resume, so I'm pretty confident that if there were a decimation I'd survive, but I think life would be very different.

ChrisMarshallNY• 5 years ago

The thing that so many folks seem to forget about programming, is that the most problematic part of the whole thing is the Requirements Phase. That's where the most serious problems occur, and they are often in the form of ticking time bombs that explode while the project is well underway. A good implementation team knows how to stop assuming, and get more information from the requirements authors.

No matter how "smart" the implementation process is, we have the principle of "GIGO" (Garbage In; Garbage Out). Automated implementation has the potential to be a garbage multiplier.

There will always be a need for highly specific, accurate and detailed specifications. If a machine does the work, and makes assumptions, then that means the input needs to be of even higher quality than we see nowadays (See "GIGO," above).

I see the evolution of "Requirement Specification Languages," which, I suspect, will look an awful lot like code.

rjsw• 5 years ago

We could feed CAD files into the same engine and get ... the 737 MAX.

WFHRenaissance3• 5 years ago

GitHub Copilot arrives from the future. Nothing human makes it out of the near-future. It's probably already too late to mount a meaningful counter-revolution. The average software engineer's status and usefulness are eroding faster than ever. There's almost nothing you can do. If you're above average, I suggest you try to start a business or con a VC before this thing blows. If you're below average... good luck, and I mean that in the most sincere way. It's possible that a career pivot is your best bet. Eastern European devs were coming for your jobs anyways, but we just took a leap instead of a step with Copilot. Sure this is a bit exaggerated, but to be clear... in this matter, the truth is that we haven’t seen anything yet.

DantesKite• 5 years ago

Sam Altman made the interesting observation that people who program might lose their jobs faster than people who move things around physically.

Which is funny when you think about it, but it does seem like it's headed in that direction.

WFHRenaissance3• 5 years ago

I think a majority of Web jobs will be automated before most "trade" workers are automated. It just makes sense. The nature of Web work itself sort of self-documents the interfaces one would need to hand to a non-technical. Trade work on the other hand is dangerous, extremely context specific, and can't be done entirely from a datacenter.

ausbah2• 5 years ago

this is really exaggerated and really really pretentious jesus

coding is only a fraction of SWEing, sure Copilot writes some code for you - but can it work with stakeholders, manage the complexity of a complicated code base, or any other part of a engineer's job

WFHRenaissance3• 5 years ago

Gathering requirements will be the product manager's job. Implementation will be the job of the top fraction of our field. Complexity can be managed by a team of humans a fraction of the size of your current team. If you're a competent engineer, you have a better future ahead of you but if you're doing routine work or templating boilerplate... it's almost over for you.

eyelovewe2• 5 years ago

And then you woke up

ipaddr• 5 years ago

Is sharing anything on github dead now? Should it be?

yudlejoza• 5 years ago

What a compiler does used to be called automatic programming in the 1960s.

lwhi• 5 years ago

Design comes first.

Implementation follows design.

I can't imagine machine 'intelligence' carrying out the former any time soon. When it does, we're done for anyway.

nsoonhui2• 5 years ago

The purpose of all this ( Intellicode), of course, is to save time.

Just wonder, how much does Intellicode actually save? Anyone has any experience?

My own take is that the saving probably isn't going to be significant, and will never be so. Because the bottleneck of programming is never caused by the speed of typing, but by the speed of thinking.

cdrini2• 5 years ago

I don't know if there are any studies on this. Personally it saves me a ton of time and mental energy. I don't have to remember the folder hierarchy when importing a function; I just reference it, ctrl+. (or whatever in your editor), and it autoimports. I don't have to remember or check whether it's SVG or Svg in this variable name. I don't have to remember what the fields of this parameter are, and so on. I don't get it wrong, build the thing, try it out, get a runtime error, and have to go back to square one to fix it. I think intellisence is useful specifically because it lets me stay in the realm of thinking about the problem/code, and avoid having to context switch to manage the details/minutia of coding.

nsoonhui2• 5 years ago

Are we talking about the same thing? You seem to be talking about intellisense ( which is there for ages in visual studio) and I am talking about intellicode, which was launched as recent as 2 years ago.

quantum_state• 5 years ago

As programming is theory building, do we seriously believe GPT3 can create quality code?

buitreVirtual• 5 years ago

While articles like this are just driven by hype and a lack of understanding, we know that a human can code spectacularly well, so in theory this same ability could be achieved or even surpassed by a true AI. I say "true AI" to distinguish that ideal from today's AIs which are great at some narrow, analytical problems but don't even come close to humans at synthesis. So while we can dismiss 95% of today's AI marketing as bullcrap, we cannot dismiss either the possibility, no matter how distant, that they could really create software.

discordance• 5 years ago

People seem to be really worried about the copyright aspect, but I don't see how that can't just addressed as a build step or a linting step.

Kind of like Turnitin, but for code. It would check similarity, and check the licence of the similar code.

JasonBlack555• 5 years ago

We've been here before ! In the 1980s there was a program called "The Last One" that was supposed to be the last program written by a human. I sometimes wonder what happened to it.

Automating programming: AI is transforming the coding of computer programs

Related Stories