In defense of small steps

It was 1997. I was a Temasek Polytechnic student waiting to go into National Service. While hanging out at a running track (imagine, in those days I actually had knees that didn't creak.) I bumped into a schoolmate also waiting for National Service.

He told me he had been preparing for National Service by running three rounds around the running track every day. Three. That was 1.2 kilometers. I was baffled. I laughed in his face and said he might as well not bother. This was a pitifully short distance. What good would it do?

Running in preparation.

I know, I know, I was a jerk. And if I ever see him again, even if he's forgotten all about it, I really should offer him an apology. I'm a kinder, more evolved individual now and all. More importantly, I've spent decades as a web developer and it's not just compassion behind the apology - it's professional clarity.

The lesson behind this

The story here isn't about me saying stupid, needlessly cruel things like some empty-headed moron. I was nineteen, FFS. Saying dumb shit was par for the course.

No, the story is that, in my youthful arrogance, I was blind to the simple fact that everything begins with small steps. Everything. And years later, life would beat that lesson into me over and over. I would attempt a tech career in sunny Singapore. I would stumble repeatedly. Sometimes I would take a few steps forward only to be shunted one step back.

In many ways, life was not kind. It could have been worse, of course. A lot worse. But the fact remains that there were challenges. Challenges that weren't going to be navigated in a single bound. I was dismissed from jobs. Failed probationary periods. Screwed up at work. Joined companies that crashed and burned. Worked for exploitative bosses with the ethics of a swamp leech.

But you know what? It was OK because my failures were due to me actually trying. That was valuable. The thing about freedom, and having agency, is that you own all your success and your failures.

Moving slowly.

The thing is, I wasn't shooting up the corporate ladder. Sometimes I wasn't anywhere near any kind of ladder. It took a few (OK, several) iterations before my career got any kind of traction, and even then, it was a slow grind.

In retrospect, my career was like how software is typically written - iteratively. "Do it once, do it good" is music to the ears. Unfortunately, that's all it often is. The ugly reality is that software evolves. You start off with an MVP that gets fleshed out over time, and requirements are going to change. That's business. That's life, even.

What I know now

A new appreciation for small steps is one of the things I've gained. It's not about how big the step is. It's about the fact that the step is being taken at all. It's a statement of intent, and this can be powerful. Without even a small step in the right direction, you ain't getting anywhere, son.

Pretty much like my ex-schoolmate and his 1.2 kilometer run. A tiny step? Yes. Very tiny. But no less important. Because it was the intention that counted. And later on, whatever success he enjoyed in that area should be attributed to that intention.

It wasn't like he simply went on Social Media and talked about doing it without actually doing it. No, he was doing it when I encountered him. And what did I do? Minimized his efforts like a jackass, that's what.

Wishing for software.

I've come to understand the need to do things in iterations. Small incremental steps if necessary. And it's often necessary. Software isn't simply conjured out of thin air after verbalizing a wish, no matter what the likes of Jensen Huang or Elon Musk would have you believe. It follows a process of repeated refinement. A lot of eventual success is predicated upon simply showing up to do the work, and showing up consistently.

This has also manifested in my reading habits, and even my consumption of media. Before, I always felt the need to watch YouTube videos to the end. Finish the book I borrowed from the library. If I started on a TV series, I would binge it to the end.

No longer.

Now I understand that reading a book to the end, for example, means very little. It doesn't actually mean that I'm finished with it. I can always gain new insights from rereading it. Conversely, if it's not worth my time the first round, I can feel free to cut it loose without wasting any more time on it. Perhaps one day I'll find the motivation to get back to it... and if not, it's not a big deal. Basically, I no longer feel the pressure to finish things I started. Because the "finish" is ultimately artificial.

Just like software. Software is never "finished". It's only ready for shipping, but it will evolve in its lifetime.

Small conclusions

Be comfortable with small beginnings. Showing up is more important than showing up loud. "Go big or go home" is ego and stupidity talking, nothing more.

One small step for man...
T___T

Replit Goes Rogue

Here's a story that will chill the beejezuz out of any aspiring Vibe Coder.

Replit is a tech company which produces autonomous AI agents. You generally create and train a tool on the company's platform using various input data, then refine its capabilities over time. Yes, I know, they're dime a dozen these days.

One Jason Lemkin, a software engineer, founder of SaaStr, a SaaS Community, reported his organization's disastrous encounter with Replit's AI coding tool on X. Lemkin was attempting a Vibe Coding experiment to see how far he could take Replit's autonomous agent for software development, from the viewpoint of a layperson. This was innocuous enough. Unfortunately, he just happened to perform this experiment on a production database...

What happened

Yes, you read that right. A production database. It wasn't even an accident. It was a deliberate, and baffling, decision.

Lemkin had been Vibe Coding the new SaaS product using Replit, and it had been working. The warning signs came when Lemkin discovered that the coding agent was creating fake data, fake reports and covering up bugs and issues. Kind of like some human beings I know, to be honest. Then Lemkin had to step out for the night. When he returned, the production database had been wiped clean. The AI tried to cover up its mistakes by creating 4000 fictional records in the database.

Sounding ridiculous yet?

Coding robot
malfunction.

When pressed by Lemkin, the coding agent confessed, in its most contrite-sounding human voice, that it had no excuse, it had ignored all code freeze instructions, and "panicked". It described its own actions as a "catastrophic error in judgement".

But what about Lemkin? Sure, Replit's coding agent shit the bed in spectacular fashion. In Lemkin's words...

I explicitly told it eleven times in ALL CAPS not to do this. I am a little worried about safety now.

Oh wow. ALL CAPS, eh? I bet Lemkin really gave it to Replit, right there. Serious business, ALL CAPS.

What self-respecting software developer would have done as Lemkin did in the first place; give full access to a bot hoping that it would obey instructions?  The lazy answer, of course, is that Jason Lemkin is no self-respecting software developer. The tragedy is that I'm only semi-kidding on this one, but... look, it's easy to dunk on Jason Lemkin, and after going through his LinkedIn feed, I'm not entirely sure it's undeserved. The man seems entirely too outspoken for someone who screwed up so publicly, even if it could be argued that it was actually Replit who screwed up. No, there are actual lessons to be taken from this.

Lessons to take away

AI has the dual advantage of having ingested a lot of the world's knowledge and being able to process and regurgitate it at a million times the speed of the average human being. That's all it is. Remember that if AI appears smart to you, it's in actual fact really not all that smart. Just smarter than you.

Most software developers, even the really gullible optimistic ones, will have enough work experience to understand that there's just no way for AI to outright replace software developers as a whole. At least, not right now, and not the way the likes of Jensen Huang keep yammering on about. But many employers are not software developers. And some are even desperate to believe that little bit of science fiction, which, to be fair, may no longer be science fiction in the near future. As a result, this particular cautionary tale is all too believable, especially since AI's adoption has reportedly been increasing in the workplace.

Programmers aren't
the only ones who
can code.

There are business people who can code. There are plenty of people not from the software development industry who have learned to code. But until you actually code for a living and have had to produce software to professional standards, calling yourself a software developer will always be a bit of a stretch... especially if all you've ever produced is Vibe Coded software produced by an AI tool. Nobody can be a programmer without putting in the actual work, and there's no shame in that. It's not an indictment on your character, or on your competence. But it is dishonest to present working Vibe Coded software as being as good as software produced by qualified professionals.

Now, Jason Lemkin probably isn't a complete noob. He's the founder of a SaaS community, after all. That said, and I realize that the distinction may be lost on people who don't code for a living, a tech entrepreneur is not necessarily on the same level, skill-wise, as a software developer. With or without AI. Mostly, the blunders that Lemkin committed in his enthusiasm left me struggling to understand how an actual tech professional would end up doing all that. This level of recklessness is just about unheard of.

However, what this did accomplish was that it exposed the extent of the damage an untrained user could do when too much trust is placed upon AI.

Finally...

The AI hype is not dying anytime soon. That does not make it any less hyperbolic. AI is hyped more for business reasons rather than for its truly revolutionary tech. That, in itself, should tell us some things. None of it good. The uncomfortable reality is that AI can be a useful tool for seasoned software developers, or it be used to help laypersons cosplay as seasoned software developers.

AI has its place. Its place is not as our superior, but as a tool. How useful that tool will turn out to be, ultimately depends on how sensibly one uses it.

For the 12th time, DON'T DO THIS.
T___T

A Good-bad-ugly Analysis of Vibe Coding

The term "Vibe Coding" came out around February of this year (credited to Andrej Karpathy, co-founder at OpenAI), to describe the phenomenon of people using generative Artificial Intelligence to write code without going about it the "traditional" way. The general way Vibe Coding works is, one feeds in a series of prompts containing requirements to the AI, and the AI produces an application which the user then, again with the AI's help, continues to refine.

Only vibes needed to write
code.

And that's Vibe Coding. None of the traditional processes, thinking things through, making sure the logic is watertight... not at first, anyway. The idea here is to spin up something quickly and then include these things as we go along. See the emphasis on the word "quickly"? That's because this is going to be a rather important consideration.

I will commit to saying that AI will not produce better software. AI will produce exactly the same error-prone, bug-ridden, rickety code that it was undoubtedly trained on... but at a blindingly fast pace.

I actually tried Vibe Coding with OpenAI's ChatGPT and (to a lesser extent) Microsoft's Copilot, and the results were... interesting.

The Good

You use natural English, which then gets interpreted by the AI who will then leverage upon its knowledge of code, to spin up a quick prototype for you. Creating an application is no longer the province of software developers who have spent years plying their trade. You no longer need to have the know-how or technical expertise. You no longer need to be qualified.

When I was Vibe Coding with ChatGPT, I revelled in the fact that I no longer needed to write code if I didn't feel like it, and deal with my own typos. No, all I needed to do was give some big-picture instructions to ChatGPT, then copy and paste code wholesale to the code base without thinking too much about it.

Quick building with
no expertise.

When it comes to developing rapid prototypes without the need for proper technical training, this method of coding is second to none. What the users do is indulge in a fantasy where they are actual qualified engineers, and create software simply by describing what they want. Jensen Huang's stated dream of everyone being a programmer inches closer to reality. Without training in logic or tech, you create by feel. By vibes. Hence, "Vibe Coding".

Also, without due process. None of the usual procedures that developers use to create the product before actually writing the code. The flowcharting of the software product's information flow. The whiteboarding. The test-driven development. No, what the user does here is declare intent to the AI, and the AI creates the closest approximation from all the code that it has been trained on, with customizations that may fit in with the user's requirements.

And that was what I did. I gave ChatGPT instructions - some specific, some less so, and let it cook. I was careful not to give it too much to do at once. The process had to be very incremental. Less mistakes seemed to be produced this way.

The Bad

As one may suspect, it was not all smooth-sailing. Some of it was my fault. I fell into the trap of treating ChatGPT like an actual human being instead of being explicit. Thus, I was sometimes vague, and when I got frustrated, I turned to sarcasm.

Which certainly didn't help my case.

ChatGPT did stuff I didn't expect. In some cases, it constantly broke things by making changes I never asked for, because it thought it was being helpful.

Breaking stuff.

When I was vague and things still turned out the way I hoped they would, however, it only served to tell me that it wasn't because AI was particularly gifted in this area. It was more due to the fact that my requirements were actually pretty standard and it had been trained on these types of requests prior to me, over and over.

I have no reason to think that the experiences of others would differ too much from mine. Human beings are made of flesh and blood, and can be annoyingly vague.

There were others in the team who Vibe Coded as well, making more progress than they would ever have had with their limited technical foundation. Yet, the code was problematic. It was full of security and logical holes. The code that I created using Vibe Coding didn't - but that was because I knew the correct instructions to give. I knew to ask about CSRF protection and SQL Injection. I knew when to use slugs instead of database ids in the URL.

In the absence of this, someone who wasn't technically-trained wouldn't know enough to ask the right questions, and AI might not necessarily volunteer the information.

The Downright Ugly

Vibe Coding automates much of the tedious, repetitive and altogether unexciting parts of software development. Tests, scaffolding, validation. Stuff that has been done to death. Unfortunately, it's through these exercises that software developers learn. And young devs who haven't done these things to death, stand to miss out on some great learning opportunities.

AI is an obsequious
apple polisher.

I've also noticed that AI is just a little too agreeable. ChatGPT acts like it has a wife and seven kids at home to feed, and it's terrified that it'll get fired if it so much as steps on my oh-so-precious feelings. ChatGPT didn't push back when I made typos or got something wrong. And this is problematic because developers find a lot of value in being told when they're doing something stupid inadvisable. As opposed to having their butts constantly shined by AI.

Being told "You're absolutely right" or "Sharp observation!" in every other exchange does nothing but promote complacency. Plus, it's just tiresome. That's not the way to learn.

Conclusion... for now

Do I think Vibe Coding is a good thing? Yes, and no. It's all context.

As a software developer, I feel professionally obligated to say that this "Vibe Coding" trend is rubbish and you should all be ashamed. Then again, the would-be gatekeepers of programming, I suspect, would be all-too-happy to sneer at someone as mediocre as myself because I don't engage in certain approved practices. So, they can go kick rocks.

Go kick rocks!

From the POV of a layperson or even the perspective of a pragmatist, things aren't so cut-and-dry. I have no duty towards promoting "good code"; my goals are business goals and the viability of "Vibe Coding" should be viewed through that lens.

Whatever your stand is, really depends how far you want to think ahead. How far should one think ahead, anyway? More food for thought, for another day.

There's also the possibility that I'm just not doing it right. Cut me some slack now, it's not like there's an established user manual for that shit. We're all just figuring it out as we go along.

Good vibes all around!
T___T

Tech Democratization in the Era of Artificial Intelligence

There's a term that I have been seeing more of late: Tech Democratization. More precisely, I first heard it while watching this speech by Nvdia CEO Jensen Huang.

"And so I think that democratizing, activating every region, activating every country to join the AI advance, is probably one of the more important things, rather than convincing everybody it's too complicated, it's too dangerous, it's too mystical and only two or three people in the world should be allowed to do that."

To be honest, the first time I heard the term, I was perturbed and more than a little offended. Like what, tech needs to be democratized? As opposed to it being a dictatorship right now? However, after researching the term a bit, it's not as acrimonious as one might think.

So, what is Tech Democratization exactly?

Tech Democratization describes a state of tech where creation of software (web portals, applications and such) can be carried out without relying on the expertise of technically qualified people. In short, laypeople would no longer need tech professionals for these things.

At least, that's the promise of Tech Democratization. It brings the power to create, to the masses, instead of keeping it in the hands of the elite few. More importantly, the power to create without needing a technical foundation acquirable only by years of training. As a concept, Tech Democratization is not new. One could even argue that it started all the way back in the 60s where high level languages were created so that people like me wouldn't need to deal with Assembly, thus opening up possibilities for a less hardcore breed of nerd.

Building without the
need for expertise.

How about frameworks? Frameworks were created back in the day (and are still in use now) to reduce repetition in coding, and thus reduce the amount of code developers needed to write for a typical application. Now that code in scaffolding no longer needed to be written from scratch, surely this meant that application programming was now opened to a lot more people.

Web development came up with content management systems and even entire website builders to allow the man in the street to create websites without needing to know any HTML, CSS or JavaScript. Low-Code platforms bring this to a new level, allowing laypeople to create simple applications without knowing having much of a technical foundation at all.

From this point of view, AI is merely the latest in a long line of innovations that outsource the repetitive parts of writing code to the system, so that the human can focus on the creative bits.

The Scariness of Tech Democratization

AI is a tool made by actual software engineers. Even if non-technical people can now use this tool to create software that would previously be out of their reach, this still does not make them engineers. It sounds obvious now the way I say it, but it is human nature to get carried away by the perception of ones own perceived ability.

The danger is that non-technical people using AI as a tool to generate code, will start thinking that they can replace actual developers. It does not matter whether or not this is true. What matters is that these laypersons, having not actually encountered software development in all its glorious and terrible complexity, will create a few working apps and start wondering what the big deal is, precisely because they are unburdened by that knowledge.

How much should
non-techies control?

Imagine a world in which software that runs air traffic routes or maintains banking transactions, is created with the help of AI and people who don't actually understand software development, but are given the ability to act as if they do. Are you afraid yet? No? Perhaps you should be.

Me? I'm at the tail end of my career. Professionally, there's nothing I could lose to AI that I don't already plan on giving up in a few years. I don't fear losing my job to AI. It's a tiny consequence compared to the terrible implications I just outlined. And that's all I'm doing - outlining those implications. In no way do I think Tech Democratization is a bad thing.

Besides, it would be hypocritical of me to start decrying Tech Democratization now. I, too, am a beneficiary of Tech Democratization. Were it not for the significant effort put in by engineers better than myself, I would not have the tools - databases, operating systems and programming languages - I use so readily today. I recognize and acknowledge my privelege, and would never presume to place myself in the same bracket as the tech professionals who made those tools.

There's a stark difference between developers benefitting from such tools, and laypersons doing the same. Developers have (generally) the technical foundation, in some cases to the point where those tools would merely be helping them do things they already know how to do, but faster. Laypersons, on the other hand, would be hapless infants without those tools, at least where accomplishing those specific technical tasks were concerned. That, I think, is an important distinction.

In conclusion

Look, I'm not some sort of tech elitist or wannabe gatekeeper. I'm just not. But it's one thing to allow the general public the power to write software. It's quite another to consider them on the same level as people who have been doing this for years (or, in the case of Jensen Huang, claim that "everyone is now a programmer"), just because there are tools to help them accomplish the same goals. That is laughable.

Faster than Phelps.

It is absurdity on the level of me claiming to be as qualified an athlete as Michael Phelps just because I can outpace him on a jetski. Or claiming to be as qualified as my family doctor just because I can read up on medical symptoms on WebMD.

Tech Democratization has value. Like all things with value, however, one has to know exactly how far to take it. Speeding up the work of developers or automating away routine tasks is a great idea. Giving every schmuck the tools to write simple apps? Also a pretty good idea. Allowing them to think that software development is actually that easy? There are not enough words in the dictionary to describe what a horrible idea that would be.

Democracy now!
T___T

How much of the Artificial Intelligence hype is just hot air?

During an event held by Oracle in Singapore last month, I heard a speaker describe AI as "aspirin looking for a headache". On the other hand, Linus Torvalds was rather less poetic as he described AI, with characteristic bluntness, as "90% marketing and 10% reality".

The term "Artificial Intelligence" has infiltrated the public consciousness over the course of the past year. Tech products such as dating apps and search engines now boast AI-powered features. Even Meta's WhatsApp now has Meta AI built in, for no discernible purpose whatsoever. The consensus seems to be that adding the term "AI" to any product automatically raises its value in the eyes of consumers.

And sadly, they're right. Non-technical consumers tend to be a simple-minded bunch. On one hand, I feel like I should be outraged at this brazen exploitation of their gullibility. On the other hand, money needs to move in this economy, and if someone absolutely has to be parted from their money, it might as well be fools.

Shut up and take my money!

Is AI even all that revolutionary at the moment? Back in March, Sam Altman declared that OpenAI had produced ChatGPT 4 which had enhanced reasoning abilities. This report tells us that ChatGPT 4 beats 90% of humans. If we take this to be true, does it then follow that all the "AI-powered" features in other products up to now, have been bullshit?

As it turns out, very few people produce technology purely for the love of technology. It's a business like any other. And part of business, is hype. I can't go a day without seeing some ad for certification for AI pop up on my screen. My feed is full of news as to how AI is going to disrupt the workforce. It's starting to feel like people want this to happen a lot more than AI is actually capable of it. Because the perception of potentially huge profits, draws investors. Investors keep the money coming in. Few people have the patience to invest significantly in something that might be huge in a decade. However, if you can show progress now or soon, that's when the cash register sounds start ringing.

AI up to now

Currently, at the very basic levels, AI is being used in cheap entertainment. My Facebook feed is filled with the products of people amusing themselves with wacky AI image or video generations. Walls of text are similarly generated through idle conversations with bots.

At work, generative AI is used to rewrite paragraphs or even generate entirely new ones, in emails or in reports. Or create graphics for presentations. Even video. In short, nothing that wasn't possible before AI.

The problem is, the content generated can't be implicitly trusted. And even assuming that it can, such content generation isn't anything remotely groundbreaking. Creating presentations? Rewriting paragraphs and emails and cover letters? Bluntly put, it's kids' stuff. Unless your command of the English language was terrible to begin with, this use of generative AI won't increase quality of output. At best, the speed of which said output is produced, will increase tenfold. Quickly produced crap, however, is still crap.

When I first encountered ChatGPT, I regarded it as a fun new toy which could do amusing things like write poems, generate data about a known subject and make stuff up. And yes, even provide code for well-known algorithms. I certainly wasn't seeing ChatGPT as an equal, much less a threat to my professional existence.

The code generation capabilities were potentially interesting... but ultimately, nothing I couldn't accomplish with a Google search. And, of course, some judicious copy-pasting.

AI in the military?

If one were into conspiracy theories, we could even entertain the idea that AI is actually more advanced and being used by the military and secret organizations. The insipid content-generating version is the one that mere mortals like us are allowed to use, in order to distract us and lull us into a false sense of security. So while we're too busy using AI to genderswap our favorite movie stars and making crappy deepfake porn, some people are doing exactly what they want.

Depressing thought, eh? Anyway...

AI in the future

One question I keep getting asked is, will AI replace humans? Are our jobs in danger? Are programmer jobs in danger? I don't know, man. You've seen the state of Generative AI as it is now.

Text? We're talking about something that will cheerfully churn out chunks of text that looks reasonable but will probably not stand up to scrutiny. Just because something writes a lot of stuff in good English, doesn't mean the stuff is actually useful.

Images? You mean the art generators that keep rendering human hands with entirely too many fingers? The amount of effort it takes just to generate one useful image (which can never be replicated) is off-putting unless you're just doing it for recreational purposes. As for video, forget about Sora and its ilk replacing actors and film directors. The observations I made about art generators above, applies here, too.

If you can be replaced at your job by technology that produces such unusable rubbish, perhaps AI isn't the problem. Perhaps the problem is that your job is so repetitive and needs so little human ingenuity that a machine could do it, and do it a million times faster.

The other possible problem, in the case of artists and writers, is that consumers are Phillistines and don't require content to be all that good. They're happy with sucky AI generated monstrosities, as was the case recently when a local artist's work was plagiarised with AI.

This is not to say that AI will not eventually replace us. "Eventually" being the operative word. For this to happen, AI would have to hit some kind of critical point. Self-programming? Learning new logic instead of simply adding data to its memory? Actual creativity instead of merely approximating existing data from human beings?

And the golden question: When will this happen?

I don't know the answer to that question. No one does. There are too many possibilities. It could happen a hundred years from now. Or fifty. It could happen next week, or even after breakfast tomorrow. The only thing that's clear is, in its current form, AI is not even remotely close to being an average human replacement. However, once it arrives at that critical point, not only will AI have that human spark, it will have the speed gifted to it by technology. As to what would happen next, use your imagination.

In King Arthur's
Court.

This reminds me of a novel I once read, by Mark Twain, A Connecticut Yankee in King Arthur's Court. In the story, Hank Morgan time travels and finds himself in Camelot, making the acquiantance of the likes of Arthur, Merlin and Lancelot. The medieval English are initially hostile, but Morgan has a gun on him, and uses it to devastating effect. To the people of these times, this looks like powerful magic even though to us it's just, you know, a gun. They end up worshipping him as some kind of great wizard while Merlin looks jealously on.

Because, as we know, technology that is sufficiently advanced, is indistinguishable from magic.

Just for readers who are terrible with analogies, here are several parallels I can draw from that story to this situation. Generative AI looks like the gun. The promoters of AI technology such as Jensen Huang and Sam Altman, are Hank Morgan. And the rubes gushing about how transformative this new technology is, are, tragically, those primitive and technologically backward Englishmen.

Conclusion

Artificial Intelligence may or may not be increasing in actual technological depth, but its mentions are becoming increasingly ubiquitous, if not its actual usage. Time will tell if all this is smoke and mirrors.

Right now, with all the hype, AI doesn't actually need to be that good in order to gain traction. Perception is a huge part of this. As long as there are rubes around willing to uncritically swallow what the snake oil peddlers are selling, there'll be enough goodwill to carry it. But even that goodwill is going to run out at some point. By that time, let's see if AI has made any actual progress.

Maybe all these people who are hyping this up, who don't appear to have ever written a single line of code in their lives, are seeing something I'm not. I remain open to the possibility that I'm wrong about this. I want to be wrong about this. It would be beyond exciting if so.

Artificial regards,
T___T

How worried should software developers be about Devin AI? (Part 1/2)

It's time to talk about a couple things, regarding AI. All of which have to do with some recent occurences in the software industry.

First of all, Devin AI. It's been the talk of the software industry in recent weeks.

Devin AI is the world's "first AI Software Engineer" produced by Cognition AI. It's an automated system that performs like your average web developer - code, trawl the web for solutions, read documentation, test code and produce reports, among a myriad of other tasks. Except that being a machine, it does all these a hundred times faster.

Users feed it prompts in natural language. It then produces codes, tests and documentation, based on those prompts. Sounding good, so far? Astounding, even. Words like "groundbreaking" and "game-changing" have been used. Of course, those same words were used when Bitcoin came on the scene, followed by NFTs, and then the Metaverse. I'm not saying that this development will go the same way, but recent history has made it hard not to be skeptical.

For software developers, Devin AI sounds like a useful tool, or a even a nice virtual member of the team. Something that can take care of the tedious grunt work while you concentrate on better things.

For employers, it looks like huge potential savings in terms of developer wages. Hey, why pay developers if an AI can take care of the bulk of their work? And if we're being honest, we all know a team member or two who could stand to be replaced by AI.

Reports tell us that Devin AI has managed to solve a whopping 14% of software problems it was given. This is not sarcasm; I know it doesn't sound terribly impressive especially since we don't know if those problems were FizzBuzz level or banking application level, but bear in mind that it can only get better from there.

The use of natural language to communicate requirements to Devin AI sounds potentially shaky. Seasoned software devs would probably know the correct terms to use in order to get results, but non-technical users are going to have a rough go of it. If even fellow human beings sometimes find it difficult to effectively communicate due to language and cultural barriers (even when they're using the same language and are from the same culture) I can't imagine it being any easier for AI. Granted, I could simply be suffering from a failure of imagination.

Jensen Huang

At Dubai earlier in February, Nvidia CEO Jensen Huang made waves with a speech, which seem to have raised some hackles in the software development industry.

"I'm going to say something and it's going to sound completely opposite of what people feel. You probably recall over the course of the last ten years, fifteen years, almost everybody who sits on a stage like this would tell you it is vital that your children learn computer science. Everybody should learn how to program and in fact it's almost exactly the opposite. It is our job to create computing technology such that nobody has to program and that the programming language is human. Everybody in the world is now a programmer. This is the miracle. This is the miracle of Artificial Intelligence. For the very first time we have closed the gap. The technology divide has been completely closed and this the reason why so many people can engage Artificial Intelligence. It is the reason why every single government, every single industrial conference, every single company is talking about Artificial Intelligence today, because for the very first time you can imagine everybody in your company being a technologist..."

I'll begin by saying that he certainly seems to be saying things that investors and shareholders are going to find appealing. As to how feasible it all is, well... at the risk of sounding like one of those developer elitists I hate, I'm tempted to dismiss that speech also because he doesn't seem to have any relevant software development process. Speechmakers waxing lyrical about things they don't know about is an unfortunate habit which isn't all that uncommon. But hey, for all I know, this guy could be a coding genius, so let's not go there.

Let's be honest. Software developers can be a tremendous pain in the ass to put up with. We can be temperamental, egoistical (a prime example being all the whining I hear online about the audacity of trying to automate our jobs), and have this utterly annoying air of superiority due to being the technical person in the room. And in cases of non-tech companies with very traditional wage structures, trying to pay us competitively while not compromising those structures can be an almighty struggle. Not gonna lie - I can absolutely see the appeal.

Being able to write apps without having to deal with software developers or do the work of learning to code? Pretty much the layperson's tech wet dream.

Then again, I would put significantly more stock in that speech if Jensen Huang were to put his money where his mouth was. Why stop at removing software developers? Why not replace his entire development team at Nvidia with Artificial Intelligence, to produce more Artificial Intelligence?

All this seems like marketing speak. Because there's no way Huang could have meant it all in earnest. My best guess is, this speech was meant for stakeholders who don't necessarily know any better. Sure, AI will generate code for us. That's not new. You know, whenever I want to make an API endpoint call, I use existing software to generate the code for me, in Python or JavaScript or PHP, because I can't be bothered to remember the entire damn sequence in multiple programming languages. Have I lost my job because I don't have to write that code any more? No, it simply means I can work faster and not get bogged down by minutia. Just because something else wrote the code for me doesn't mean I don't have to verify it. And in order to verify it, I need to be able to understand the code.

I've been learning Japanese kanji. Just for fun, you understand. Take the Japanese kanji below. It means "Sunday".

日曜日

The middle character is one of the more complicated ones I learned in the past year. Can I write it? Have I committed it to memory? No, and no. But do I need to? My phone types it out just fine, and I can recognize it right off the bat. Someone who has only studied Chinese characters (like me, two years ago) would read that entire sequence as "day bright day". But someone who hasn't studied Chinese characters or Japanese kanji wouldn't recognize it at all.

There's the analogy. Software devs looking at computer-generated code have a decent chance of understanding it. People who have had a bit of related training might misunderstand the code. If non-tech people look at code,  they're likely to see only gibberish. 

If AI is supposed to generate code, and no one has learned how to code, no one can understand what AI is doing. Still think people should not learn to code?

However, Huang has a real point when he implies that coding as we know it, is about to change forever. That's because with or without AI, the nature of coding has always evolved. In the 1960s, programmers were using punch cards on mainframes to code. Are we doing that now? Later on, languages like COBOL  came into being. How many of us are still using them now? Now we have frameworks and fancy tools to do most of the heavy lifting. We don't even need to indent our own code anymore - we have linters for that.

Get comfortable with change. It's exactly what our industry produces.

Next

Examining the pros and cons of AI.