vickychijwani

With AI you can be either a centaur or a reverse centaur - as long as you delegate only the implementation but maintain control over the direction, outcome and taste. This means only 1-2 tasks at a time.

When you begin to delegate high-level thinking to the machine, which is inevitable with “many tasks, all at once” (the alternative is burnout like the article describes), that’s when you transform into the reverse centaur.

I’m happy to engage in meaningful discussion; this is just sci-fi exaggeration (Wall-E?).

Sci-fi has tainted our collective imagination of a tech-forward future - it doesn’t have to be that way.

Two things can be true - yes, AI can benefit mega-corporations AND drive spectacular human progress at the same time. Is all of it “good” by my ethical standards? No. Is it still valuable? Yes, immensely.

If you know something to be true, without knowing why, does that reduce the value of said truth to zero? And what if that truth has positive effects on your ability to invent and leverage technology, treat/cure disease, etc? Would you still say it is of no value and people are “falling for it”?

This is called faith, and as strange as it may sound, in a few decades this kind of faith in the machine will be commonplace, because it will prove its utility over and over and over again, and save mountains of human effort in the process. In 3-7 years we’ll start to see major practical applications from AI-led advances.

As a second-order effect we may even see the rise of an actual religion around AI. I’m not saying these are all positive effects, but it’s what I see happening - and like all technological revolutions there will be a huge number of positive effects alongside some deeply disturbing developments.

Indeed. I would say that’s the healthy way to approach any such pursuit. Else the economics taints everything if you let it.

There is still emotional and social value. The AI is a tool in the hands of a competent operator.

You’re blinded by your hate for corporations, if you cannot see value here. The actual human who’s working on it responds, and you can’t help but make it about the AI that you believe is somehow only benefiting the corporation.

Given your preference you really should avoid that test. Science barely understands the human body, despite all the overconfidence and pretense. We do not have “maximum” information about our bodies in any meaningful sense; only half-knowledge that often backfires spectacularly. That is why there is so much subjectivity in doctors’ opinions and such.

“I don’t think it would be long” is an age-old planning fallacy in this field, AI hasn’t changed that. It is certainly easy to clone a UI today because an agent has a target to iterate against, but cloning all the behaviours of the software in question is going to prove to be much more difficult.

These are decades-old software where even the inconsistencies, quirks and bugs are now considered features (because users have come to rely upon them) - a full detailed specification itself would be very complicated. My bet is this OSS-clone hype will last all of 6 months.

But this is not even the easy 80%. More like a skin-deep “reimplementation”.

As an Indian the hypocrisy is mind-bending, and didn’t begin with Trump. This is just the latest, most transparent example.

India is pressured constantly for making inconvenient decisions in our national interest, when that is the only way the US has ever operated historically.

Yeah, HN is unfairly dunking on a cool project that actually adds tremendous value for normal folks who don’t spend all their time at a computer.

Wrong target audience I guess, but I’m surprised by the sheer short-sightedness on display here.

This isn’t tin foil hat, it’s standard corporate strategy - the more of the customer relationship you can own, the better your long-term retention and growth will be.

Only for folks who already know what they’re doing.

What I said will only make sense if you take yourself out of your current context and think entirely from the perspective of someone who knows little-to-nothing about ML.

It’s the same mistake folks on HN made when Dropbox launched, drawing comparisons to rsync and other Unix tools as if they were somehow equivalent.

Tons. For example, most web and mobile app developers won’t know where to start with making a bespoke model (and would likely have no interest in making one), but they will have lots of usecases for a classifier.

We’re not far off from the point where a 30B parameter model could do that and run on not-too-expensive hardware. See recent Qwen releases for example and extrapolate the current rate of progress from there.

I see. Can you say more about this? What’s the trade-off of removing it?