I'm sure you have seen Github's (Microsoft's) Copilot by now, which claims to help the programmer by generating randomly broken code which everyone hails as a total game changer that will forever change what it means to program. Everyone seems to agree that it is the coolest thing since garbage collection, which makes me wonder: am I retarded for seeing it more as a nuisance? How does cleaning up after randomly generated mess supposed to help? It looks to be more work than just writing the code myself. Is it just hyped by retards who can't program anyway or am I mentally challenged?
Metaprogramming with self trained ai isn't new and copilot is garbage fire. Trained off a webshit dataset and closed source tooling for development.
Is it just hyped by retards who can't program anyway
Javascript code monkeys at best.
am I mentally challenged?
Midwit blinded by the retards. Tools are useful not toys, copilot is a metaprogramming toy that spits out entropy from a mostly garbage dataset.
>>1
At best this seems like a heuristic designed to mitigate the symptoms of languages without sufficient means of abstraction to avoid boilerplate, and overly complex APIs. At worst this is yet another mechanism to deskill the profession, encouraging random guessing, and exhaustive testing of every possibility by an army of fungible cogs rather than understanding, and intelligent design by a select few.
Encouraging random guessing, and exhaustive testing of every possibility by an army of fungible cogs.
It's also worth noting that unlike the Jacquard loom what this mechanism will do is decrease efficiency in terms of labor. Perhaps in the near future we will see programing by Amazon Mechanical Turk.
So is this some grand conspiracy from Microsoft to make programming less fun?
So is this some grand conspiracy from Microsoft to make programming less fun?
The following explanation may be more to your tastes:
The market has noticed that there are insufficient programmers to meet demand. Microsoft is a profit maximizing company seeking to provide a service as requested by the market. Microsoft (along with every other company) is not in a position to make people capable of programming, as this would require proper nutrition, and education for children, reducing contaminants and pollution, reducing spectacle and distractions, and marinating these youngsters in the field, etc. Not to mention that resolving these issues could increase their cost of business, or on their officers substantially through taxes!
So barring the possibility of making complete human beings the world has seen 1001 ways to reduce programmer cost by attempting to offload onto armies of apes, which would otherwise be few men. These replaceable parts don't need to read documentation, or specification, only examples which they can imitate without thinking. They get by fine because almost everything they're doing has been done many times before; typically their only challenge is locating a new example to copy. They meet performance criteria since their solution passes an exhaustive testsuite in a timely manner, and because no one cares if the software is well designed and thought out. Unfortunately programming is a thinking game, you have to understand, and contemplate to find ``The Right Thing'' in both interface and implementation. The result is a cancerous growth of complexity and ``The Wrong Thing'' so as to make thinking and ``The Right Thing'' nearly impossible.
With this growth we are presented with two options, let men rebuild, or more monkeys. We repeat from the beginning. This is how things have always been here.
The market has noticed that there are insufficient programmers to meet demand.
I'm not buying this. Every company these days have insane interviewing processes that span multiple days and involve hours of unpaid work. Clearly there's such an abundance of programmers that it is cheaper for companies to find the "perfect" candidate than having to take on someone less-than-perfect and invest in training them. They are certainly in the position to make programmers live up to their own expectations, but right now they are in a very comfortable situation where all the costs associated with training are pushed over to the state/society and the individual.
>>7
The whole point of monkey labor is that monkeys are cheap and plentiful. You've got a mile of monkey applicants, and your objective is to find those best capable of opening bananas. Even safety critical systems are now often made by monkeys a la Boeing 737 MAX.
>>1
Isn't genetic programming [1] already capable of automatically generating code? Lisp is particularly suitable for implementing genetic programming algorithms.