In March 2023, a new AI research lab called Cognition Labs released a video on YouTube introducing their newest product; Devin, the so-called “first autonomous AI software engineer”. The preview of the product produced quite a sensation, mainly because Devin’s abilities were shown to be on par with that of a software engineer.
Equipped with the ability to write, edit, and debug code, Devin can perform coding tasks with greater accuracy than other AI models. What truly made Devin the talk of the town was that it was able to accept gigs on Upwork and earn money by completing software engineering tasks. As always with every AI product, the talk of ‘replacement’ also took center stage. Videos with titles like “AI is going to REPLACE Software Engineers” were flooding our timelines. Now that three months have passed, we can ask the question “Where is Devin now?” and the answer we get is…crickets. Once again, the AI market has (probably intentionally) fallen victim to the specter of hype.
When optimism outpaces reality...
Devin promised to deliver quite a lot. Scott Wu, CEO of Cognition Labs, claimed Devin could perform at the same level as human software engineers. This assertion framed Devin not merely as a technological breakthrough but as a possible substitute for human professionals in this specialized field. Any software engineer's initial reaction to the preview video of Devin would be skepticism. And that skepticism is warranted. However, most people aren’t software engineers, and what we got was tweets like this:
But is it true? Can Devin perform at the level of human software engineers? Well, no. Don’t take my word for it. It was actually an expose done by a YouTube channel called Internet of Bugs that brought this fact to light. The creator of the video evaluated the claim that Devin was able to take gigs on Upwork and complete them efficiently. The video exposed many issues in the process that ultimately showed that the company was outright lying about (or at least misrepresenting) Devin’s abilities.
Another reason why we should maybe question Devin’s true abilities is the fact that it’s still not publicly accessible. There is a “preview” version of Devin that is supposedly open to use. However, to access the preview version, one must first log in or create an account and get approval from the company - which is the opposite of what a preview version should be. I don’t know about you, but the whole process reeks of a data collection scheme.
Premature Claims…again
The lesson here is quite obvious. Be skeptical, don’t trust the hype, et cetra et cetra. But lessons aside, this points to an interesting market phenomenon that’s becoming increasingly prevalent thanks to AI. I mentioned in the intro of this article that the AI market is falling victim to hype intentionally. Let’s see why.
Hype-Economics
Clout is everywhere. It directly or indirectly affects different aspects of our lives and by extension, our economy. With the introduction of generative AI, a new trend is starting to take hold. Some obscure startup pops up out of nowhere promoting an AI product that is a “game changer”. They then launch an extensive marketing campaign promoting said game changer. The general public eats this up mainly thanks to “tech influencers” producing clickbait content like “AI is gonna replace [insert profession]”. However, after a few months, the product and its company either fade back into obscurity or fail to deliver on their lofty promises.
While this may look like a disastrous business model, it’s trending for a reason. The crux of this scheme lies in generating the hype. Generating hype around a startup will artificially inflate its market value and as a result, attract funding from venture capitalists. Furthermore, as the company grows (artificially), its likelihood of being acquired by tech giants such as Google will increase. Throughout this process, the founders make a lot of money. Seeing this, we can understand why AI startups often overestimate their products.
As a conclusion, I’ll say this; Devin is probably a pretty cool tool. Developers can stand to benefit from its features. However, the phenomenon of hype economics is drastically limiting our interaction and use of AI tools. Although the general public says so, the relationship between humanity and AI shouldn’t be antagonistic. Many AI products are genuinely helping people in various sectors without being marketed as replacers. While the incentives of hype-driven strategies are great, startups should also consider transparency and responsible marketing…just for a change.
