Matt Welsh is the CEO and co-founder of Fixie.ai, an AI-powered software distribution company founded by a team of Google and Apple. “I think the conventional idea of ’writing a program’ is about to disappear,” he opines in the January article. ACM Communications“and in fact, for all but the very specialized applications, most software as we know it will be replaced by artificial intelligence systems that are trained instead of programmed.”
His essay is titled “The end of programming” and predicts a future “Programming will be obsolete.”
In situations where a “simple” program is needed (after all, not everything should require a model of hundreds of billions of parameters running on a bunch of GPUs), those programs themselves will be generated. by an AI instead of hand-coded. ….with humans relegated to, at best, a supervisory role…. I’m not just talking about things like Github’s CoPilot replacing programmers. I’m talking about replacing the whole concept of writing programs with training models. In the future, computer science students will not need to learn such mundane skills as adding a node to a binary tree or coding in C++. That kind of education will be old-fashioned, like teaching engineering students how to use a slide rule.
Engineers of the future, with a few keystrokes, will activate an instance of a four-quintillion-parameter model that already encodes the full extent of human knowledge (and then some), ready to perform any task required by the machine. Most of the brainwork of getting the machine to do what you want will be finding the right examples, the right training data, and the right ways to evaluate the training process. Powerful enough models capable of generalizing through few shot learning will require only a few good examples of the task at hand. Massive human-selected data sets will no longer be needed in most cases, and most people “training” an AI model won’t be running gradient descent loops in PyTorch, or anything like that. They will be teaching by example, and the machine will do the rest.
In this new computing, if we call it computing at all, the machines will be so powerful and already know how to do so many things that the field will seem less like an engineering endeavor and more educational; namely, how to better educate the machine, similar to the science of how to best educate children in school. Unlike (human) children, however, these AI systems will fly our planes, operate our power grids, and possibly even rule entire countries. I would argue that the vast majority of classical CS becomes irrelevant when our focus turns back to teaching intelligent machines instead of directly programming them. Programming, in the conventional sense, will in fact be dead….
We are moving rapidly toward a world where the fundamental building blocks of computing are temperamental, mysterious, and adaptable agents… This change in the underlying definition of computing presents great opportunity and many enormous risks. However, I think it is time to accept that this is a very probable future and develop our thinking accordingly, rather than just sitting here waiting for the meteor to fall.
“I think the debate right now is mainly around the measure to which these AI models are going to revolutionize the field,” Welsh says in a video interview. “It’s more a matter of degree than whether it’s going to happen…
“I think we’re going to shift from a world where people mostly write programs by hand to a world where we teach AI models how to do the things we want them to do… It’s starting to feel more like a field than focuses on AI education and maybe even AI psychiatry. To solve these problems, you can’t just assume that people are going to write the code by hand.”
#AIpowered #software #delivery #company #predicts #programming #Slashdot