Is the end of programming near?
If you ask Matt Welsh he would say yes. As Richard McManus wrote on The New Stack, Welsh is a former Harvard computer science professor who spoke at a virtual meeting of the Chicago Association for Computing Machinery (ACM) and explained his thesis that ChatGPT and GitHub Copilot represent the beginning of the end of the Programming.
Welsh joined us at The New Stack Makers to discuss his perspectives on the end of programming and answer questions about the future of computing, distributed computing and more.
Welsh is now the founder of Fixie.ai, a platform they are building to allow companies to build applications based on large language models to extend them with various capabilities.
Programming language design has had one goal for 40 to 50 years. Make it easier to write programs, Welsh said in the interview.
Still, programming languages are complex, Welsh said. And no amount of work will make it easy.
“It seems unlikely to me that any amount of work to improve type systems or syntax or any of these debugging measures will suddenly crack this nut and suddenly make programming easy,” Welsh said. “We’ve been at this for a while. It’s not improving. So this is where I think there has to be some kind of quantum leap to stop programming to talk to computers and teach them.”
It’s like when few people could read books, for example.
“Well, if computers are, shall we say, democratized because now you don’t have to be some wizard in a tower who understands how to write Rust code to instruct a machine, then that dynamic becomes complete change.” Welsh said. “Anyone will make it. And I actually think that’s a good thing. You know, there are all kinds of people in the world and places in the world that could benefit from computers that just don’t have access to it because the skill level, the skills required, are just way too high.”
Computer science has always been about people taking a problem and turning it into instructions for a machine, Welsh said. That is the definition of computer science. It is the art of science to map problems to what machines can do. Now that models are getting bigger, it’s no longer an x86 CPU executing the machine instructions.
“So now your core isn’t an x86 CPU executing machine instructions,” Welsh said. “It’s an AI model that solves problems. And, you know, operating and working in many ways as a human might.”
Group Created with Sketch.
Alex Williams is the Founder and Editor-in-Chief of The New Stack. He is a veteran technology journalist who has worked at TechCrunch, SiliconAngle and what is now known as ReadWrite. Alex has been a journalist since the late 1980s and started…
Read more from Alex Williams