W.hen digital computers were invented, the first task was to instruct them to do what we wanted. The problem was that the machines didn’t understand English: they only knew ones and zeros. You could program them with long sequences of these two digits, and if you got the sequence right, the machines would do whatever you wanted. But life is too short to compose infinite strings of ones and zeros, so we began designing programming languages that would allow us to express our desires in a human-readable form that could then be translated (using software called a “compiler”) into terms that machines could understand and obey.
Over the next 60 years or so, these programming languages – with names like Fortran, Basic, Algol, COBOL, PL/1, LISP, C, C++, Python – proliferated like rabbits, so that there are now many hundreds, perhaps even thousands of them. In any case, it takes quite a while to scroll down to the end of the Wikipedia page listing them. Some are very specialized, others more general, and over the years, programmers created libraries of code snippets (called subroutines) for common tasks (search and sorting, for example) that you might incorporate when writing a particular program.
Thus, for more than half a century there evolved an arcane and exclusive priesthood of people who mastered one or more of these specialized languages and were able to make computers do their bidding. Belonging to the priesthood gave one a heady feeling of absolute power. Remember that a set of pixels can be programmed in software to move infinitely in a circle, for example, and they will continue to do so forever if left that way. They don’t need fuel or food and will never complain. “In that sense,” I once wrote while writing a history of this technology, “being a programmer is like being Napoleon before Moscow retreated. “Software is the only medium in which the limits are exclusively those set by the imagination.”
So when large language models (LLMs) like ChatGPT emerged, many people were shocked to discover that these machines could not only compose coherent sentences in English, but also They could also write computer programs.! Instead of having to master the byzantine intricacies of C++ or Python in order to converse with the machine, you could explain to it what you wanted it to do and it would spit out the necessary code. You could program the machine in plain English!
How was this possible? Essentially because, in its training phase, the machine has absorbed a large amount of published computer code, just as it has also absorbed practically all the tests that have been published. And although the computer code it produces often has flaws, they can often be fixed in successive iterations. The technology is already pretty good, which is why programmers have been early adopters. a kind of “copilot”. And it will constantly improve.
So are we seeing the decline of the software priesthood, like some of the most apocalyptic reactions to LLM claim? Personally I doubt it, if only because we always overestimate the short-term impact of technological change, while underestimating its long-term effects. What these AI “copilots” really do is take the drudgery out of programming, freeing up those who understand the software to do more interesting and productive things.
When GitHub, the Microsoft-owned developer repository, quizzed more than 2,000 software professionals about the technology, the results supported that view: 88% said it made them more productive; 59% said it made work less frustrating; 74% said it had allowed them to focus on “more satisfying work”; 96% found that it made them faster when performing repetitive tasks; and 77% said they now spent less time searching. This is not the image of Armageddon, but of something more positive.
And AI copilots are already starting to change the way programming is taught. Most introductory computer science courses tended to focus on code syntax and getting programs to run, and while knowing how to read and write code is still essential, testing and debugging should now be taught more. explicit. Academics are finding that having students use AI tools frees up time “for teaching.” high level thinking – for example, how to design software, what is the right problem to solve and what are the solutions? “Students can spend more time on the optimization, ethical issues, and usability of a system rather than focusing on the syntax of the code.”
Mario Fusco, a great software developer, once said: “The code you write makes you a programmer. The code you remove makes you a good one. Code you don’t have to write makes you great.” So maybe, for once, AI has achieved something.
after newsletter promotion
what i have been reading
Student disunity
very perceptive blog entryUniversities As Factories, by economist Branko Milanovic on the Columbia protests.
short is sweet
The tyranny of content algorithms: A succinct essay on the importance of quality over quantity written by Silicon Valley veteran Om Malik.
Where Walter went wrong
What Elon Musk’s Favorite Game Tells Us About Him is Dave Karpf’s Title scorching criticism from Musk biographer Walter Isaacson.
