New ask Hacker News story: Ask HN: Are you worried about generative programming AI models?
Ask HN: Are you worried about generative programming AI models?
2 by spacephysics | 0 comments on Hacker News.
We're starting to see more sophisticated uses of code generated by AI models that go beyond the typical IDE intelli-sense. Recently there was a HN post about an AI model [0] (believe a NN) that wrote a bunch of CRUD Python code given human sentences as input. Though the example seems simple, I think it's implications are far reaching. Personally, seeing the example has shortened my estimated timeline for "deep" code generation by a substantial amount. Naturally I wondered how soon until basic CRUD applications, and other low-hanging software development fruit are picked by AI driven software. I think the average programmer is safe for the next 5 years, but soon after only highly domain-specific use cases for human programming will be cost effective for businesses. CRUD apps, static websites, and perhaps more complex applications with some human programming, will greatly reduce the needed headcount of programmers in a typical business. I see this similarly to IT, where initially IT teams were much larger, however more recently with the advent of DevOps processes and technology, a small team of IT-centric developers can reliably handle a medium-sized company. Likewise, an easy value proposition will be the aforementioned AI models as products to reduce development cost for simpler applications. Should we begin to prepare for this wave now by learning some tangential sub-field to retain similar magnitude of job opportunities? If so, what would be some examples? Obviously we should be constantly learning, its a necessity in the field. My point isn't to fear-monger, rather contemplate if this wave could be larger than the typical language/technology advancement, both in magnitude, and in time of adoption. [0] https://ift.tt/3dY8nXc
2 by spacephysics | 0 comments on Hacker News.
We're starting to see more sophisticated uses of code generated by AI models that go beyond the typical IDE intelli-sense. Recently there was a HN post about an AI model [0] (believe a NN) that wrote a bunch of CRUD Python code given human sentences as input. Though the example seems simple, I think it's implications are far reaching. Personally, seeing the example has shortened my estimated timeline for "deep" code generation by a substantial amount. Naturally I wondered how soon until basic CRUD applications, and other low-hanging software development fruit are picked by AI driven software. I think the average programmer is safe for the next 5 years, but soon after only highly domain-specific use cases for human programming will be cost effective for businesses. CRUD apps, static websites, and perhaps more complex applications with some human programming, will greatly reduce the needed headcount of programmers in a typical business. I see this similarly to IT, where initially IT teams were much larger, however more recently with the advent of DevOps processes and technology, a small team of IT-centric developers can reliably handle a medium-sized company. Likewise, an easy value proposition will be the aforementioned AI models as products to reduce development cost for simpler applications. Should we begin to prepare for this wave now by learning some tangential sub-field to retain similar magnitude of job opportunities? If so, what would be some examples? Obviously we should be constantly learning, its a necessity in the field. My point isn't to fear-monger, rather contemplate if this wave could be larger than the typical language/technology advancement, both in magnitude, and in time of adoption. [0] https://ift.tt/3dY8nXc
Comments
Post a Comment