@xmanplez@jessesingal this is true but at some point it won't matter
these companies have billions of dollars that they can spend to create bespoke training data for whatever is required.
What’s the first bit of software you remember changing just because you COULD?
HTML on a personal page? Game config files? IRC scripts? Winamp skins? Cracking open some horrible PHP file on a shared host?
That feeling of “wait... I can change this?” got me good😂
Contemporary artificial intelligence is to cognition as artificial sweeteners are to good nutrition: full of misleading hype, distracting, and when misunderstood and misused, dangerous to one's well-being.
The people being like "oh just a json classifier" are honestly pretty stupid.
Give Jev weird, high-cardinality decision problems you couldn’t plausibly train a dedicated classifier for. Or think about every startup that needs useful decision signal immediately, before it has a dataset to train one
"why not use an llm" because they're slow as fuck and costly
A gentle reminder that contemporary neural networks are only an abstraction built upon Boolean logic
And that said neural networks are only an echo of a whisper of what organic neurons are.
Very little about software engineering has changed over past last three months.
A great deal has changed about coding, not unlike when we saw the rise of high order programming languages and compilers, the difference today being that the number of developers is far larger and distribution channels are such that the velocity and breadth of change is far greater.
The entire history of software engineering is one of raising the level of abstraction.