Devin AI Will this AI Replace Programmers?

Today this unholy news became popular, that an “AI Software Engineer” with the name Devin, made by Cognition Labs , will be available on the market. And, of course, it provoked a strong reaction among the software professional community. Reactions ranging from cynicism to fear to disappointment abound in the comments on the related videos.

While we are still far from having a self-aware AI , what we do have is a product with promising features that threaten to turn the technology industry upside down.

Like what features? According to its creators, Devin can :

• Use unfamiliar technologies
• Create and deploy applications
• Automatically find bugs
• Train your own AI models
• Contribute to mature repositories
• Solve real projects on Upwork (much more than solve interview code problems )

Who is behind Devin?

Behind Devin is Cognition labs, a relatively new company that had been kept behind the scenes until yesterday. Their Twitter account is less than two days old at the time of writing (although it already accumulates over 86k followers).

Who are the members of Cognition labs?

There are about 10 members, between them totaling more than 10 gold medals at the International Computer Olympiad and whose members seem to have been involved in AI-related projects at companies like Google, DeepMind and Scale AI. So one can speak of a group with some level of experience.

Who is funding the project?

So far Cognitive Labs claims to have raised about 21 MDD in a round led by Founders Fund, most notably investor Peter Thiel, former head of Paypal.

What’s going to happen to the programmers - are they at risk?

Well, it is difficult to know for sure, given the lack of information. However, I propose a couple of likely scenarios:

• Devin will replace the programmers and revolutionize the industry.
• Devin will not rise to the occasion and will fall into oblivion.

Devin will replace the programmers and take their jobs.

Who do you think would win in a battle to produce more books, 1000 ancient Egyptian scribes or a photocopier hooked up to a computer?

I am of the opinion that the second option, simply the capabilities of the printer to produce pages faster, with fewer errors and in a fraction of the time, are beyond the capabilities of the scribes. While it is true that a person is required to set up and use the printer, that does not imply that scribes stand a chance. Similarly, while programmers will still be required and more jobs will be “created” as is usually claimed, the real question is: how many jobs will be lost for every new job that is created?.

Let’s continue with this thought experiment, let’s assume, for a moment, that Devin does exactly what he says. Since Devin is a tool, it still requires input from a human being to function, so the “output”, i.e. its performance, will depend on how good its “input”, i.e. the person giving it the instructions, is.

This means that the profession of prompt engineer will become a reality, and with the consequence that those people with superior abstraction, writing and lexical abilities will become practically gods of this new utopia (or dystopia, depending on how you see it).

In addition, all those professions in charge of perfecting AI models will become professions with a very high demand, since any company that wants to compete and stay afloat will need one.

New opportunities after the end of programming

Getting a job if you are new in the industry will be quite complicated, but at the same time the barrier to create digital businesses will decrease and it will be more important to identify and solve problems. Becoming a solopreneur or freelancer will be incredibly easy, as the technical skills required will be significantly reduced and you may be better off aiming in that direction.

How likely do I see this scenario? The truth is that I consider it a pretty unlikely scenario.

Devin is just temporary hype and will be forgotten soon.

The second possibility is that nothing happens. Yes, nothing.

It has happened multiple times in the past; every great forgotten idea starts with the promise of revolutionizing the world, as it happened with no-code or Dreamweaver (A WYSIWYG editor to create websites from 2007). Both were technologies that from one moment to the next threatened to change the world of technology forever, and failed to do so.

Remember Agent Swarm AI ? It also generated a lot of expectations but in the end it didn’t crystallize into anything.

So far Copilot, ChatGPT and probably Devin have been nothing more than “google searches” on steroids, they can give you the information you request directly, but it’s not necessarily real and fail-safe. The iterative process offered by this system improves quite a bit on the usual question-answer dynamic, but it is still immune to falling into loops due to problems it can’t solve or producing something totally different from what you ask for, for now we need to test it a lot more.

In addition to the above, there is another aspect that we are missing.

The financial aspect of Artificial Intelligence.

Let me describe a situation:

1. A group of entrepreneurs come up with a novel product.
2. It goes viral and everyone talks about how it will forever change our everyday life.
3. The creators receive seed capital from many, many investors.
4. It ends up becoming just another tool and everything stays the same.

Does it sound like anything like that? Maybe they are just a group of people looking to become millionaires and we’re in the middle of an AI bubble .

Consider that, in the world of money, appearances play a major role. Right now, only a few people who have had access to Devin are raving about the tool. Sounds too good, doesn’t it? This could all be a marketing campaign to inflate expectations, attract investors and make “easy” money.

But didn’t Peter Thiel just invest money? Yes, but that doesn’t mean anything. If you are an investor, you don’t care if the project exists for a month, two months, a decade or two, the only important thing is that it is profitable enough for you according to your own investment criteria. In other words: What does it matter if the project dies if I can turn 21 MDD into 40 MDD in less than a year?

It strikes me that the way to request access to Devin is through a google form instead of using an App developed with the same model, which would be the perfect cover letter for this technology. After all, it’s not like it’s outside the capabilities of their model, it’s not an application with 1000 microservices, load balancing, streaming and a recommendation engine, it’s a plain simple and common registration form.

I consider this scenario more likely, but let’s wait a year or two and see together if Devin lives up to its promise. My bet is that Devin won’t replace programmers.