How do you train the AI...when Stack Overflow goes down?
They've been trying to hire software engineers to write code as training data for them. I don't know how successful it has been, but I admit I would write code for them if they paid me enough.
I was working on a twenty-year-old Java project which had very few unit tests last year. In a couple of days with Copilot I added tests for a couple of hundred classes; for the simple ones it generated the entire test and I just needed to run it, for more complex classes it needed more help since it didn't understand things like not trying to mock final classes. The tests also found a handful of bugs that had been in the code for years so I fixed them.
What kind of code coverage did you get from your tests?
I think it would be more interesting to look at what Copilot did for those 20M users.
While true, Microsoft has a long tradition of only showing the numbers that make themselves look good..
As an example, when Windows Phone 7 was released, they cited several millions of downloads of their IDE, instead of declaring the total number of phones sold. That is why we are up to WindowsPhone 17 today.
Then that organic chemist needs to go back to school and revisit the topic of chiral molecules...
Ok, I'll let you email the authors and tell them they are wrong. I'm sure they'll be happy to hear from such a smart person as you.
Pointers aren't required for most purposes. They're often just an optimization, frequently a questionable optimization. It's true that C pretty much requires pointers, but in C++ references can generally be substituted with greater clarity. Pointers are almost never used in Java (are they ever?), and certainly not in Python. Or many other languages I could name. (Yeah, they still exist "under the hood", but that's not the point of an exam of early or intermediate programming skill.) For that matter check out D https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fdlang.org%2F . That's a language that would be my favorite if they had a better way to document your code (last I checked Doxygen didn't do a good job) and it it had a slightly better library. (As it is I currently prefer C++ except for stuff that's heavy in unicode, where I'll switch to Python.)
C is actually easier than Java, though doing anything complex is harder.
A C++ subset doesn't have to be any harder than Java, and can be at least as capable. (The problem would be documenting it...and for students, learning to use only the subset.)
"Did Craigslist Really Kill the Newspaper Industry?"..."Did Craigslist drive the downfall of print classifieds?"
Those are actually two very different questions, and should not be grouped together (unless it can be shown that one caused the other).
If you have a coverage checker, the LLM can iterate until it covered everything. That's not hard to do.
Yes, so solve the simple problems before they start pretending they can solve the more complex problems.
So is the price of energy to power AI dwarfed by these billion-dollar payrolls,
That's a good question.
Science has advanced, but society is still essentially unchanged, and without a massive genetic redesign it probably canâ(TM)t change, which is probably a good thing.
I don't think that's really true, though; it's just a matter of education.
That's not really a scientific hypothesis because you're not attempting to gain knowledge about nature
False.
To put it another way, positing an outcome and then observing that outcome, is not science. This was the logical positivism that Popper criticized. Positing a reason for the outcome is science, and even if the experiment reliably demonstrates your prediction, it has not proved any hypothesis for the reason the experiment reliably conforms with the prediction.
OK, that's a better way to put it.
A company is known by the men it keeps.
