Sadly, Kagi uses a Cloudflare turnstile that's set to paranodial mode, so my Firefox on Linux is rejected. I suspect it's because the turnstile wants to use some WebGPU APIs to fingerprint my browser, and those APIs don't exist.
IMO, this captures well the essense of AI: it's an alien swarm intelligence that solves problems with brute force and pattern matching. For comparison, we prefer simple and elegant solutions, partly because our time and attention scope is very limited, but mainly because it rhymes with our nature.
I think we like simple and elegant solutions because we understand the perils of leaky abstractions. LLMs benefit hugely from this also; short scripts that work well can save them hours of faffing about trying to discover how to do something. I imagine if they were able, they too would prefer elegant solutions
Because AI's language isn't a language at all, but its imitation. Imagine you're given a map skillfully created by someone who has seen many maps, but never bothered to understand their meaning. How much time will you waste before realising that this map is an imitation?
To appreciate music and art you need a sense of harmony. An alien species with aversion to harmony would produce similarly disgusting civilization: its music will be noise and its art will be chaos.
The unspoken question here is "can a species with an aversion to harmony form a civilization, something requiring some form of strong and extended cooperation?".
The assumption was that some engineers knew how each part of the system worked, or at least the system was comprehensible because it was created by humans for other humans. There were comments explaining difficult places, etc. With AI this assumption is broken. Now nobody understands how things work, and it cannot be understood in principle because the system is a mountain of slop created by some alien intelligence.
I'm a staff engineer on team of 20+ engineers working in a single codebase. We're all pretty heavy users of AI, like most engineers today.
I worry about a lack of understanding creating risk, and we are creating tools and conventions to address this risk.
But I don't yet see evidence that we're losing touch with what we're building.. yes, code is shipping faster, but I'm honestly just not seeing this "sleep-walk-engineering" issue in reality.
Code isn't shipping so fast that the engineer doesn't understand the intent and general implementation details of the PR.
PRs are required to be small (400ish lines), and have tests, and good PR descriptions must be present. And two humans must approve the PR.
The time it takes to understand, verify and validate the change is fast, often as fast as or faster than the time it takes for AI to implement.
As it stands right now, our level of understanding, PLUS our new verification tools... is all we seem to need.
We're not like the hyperscalers, but we're considering how some of their approaches might fit into our workflow.
And much like Anthropic, we aren't seeing bugs increase with the rate of code increase.
Engineers are still able to understand enough of what they need to.
Well, the fancy calculators do. Yes, they still need guidance to keep from going off the rails. But, at least from personal anecdotal experience, I need to go spelunking into the codebase less and less over time. The agents and models really did just start to become that good.
FDA approval of the first GLP-1 agonist was in 2005 and many diabetics have been taking these drugs for close to 20 years, so we have a pretty good picture of what long term side effects are: mostly GI symptoms, and you can have problems with muscle loss and bone weakening if you lose a ton of wieght. otherwise they are well tolerated.
Been on one for a few years now, it tempered my persistent auto-immune issues. Short term side effects are barely present for me. This stuff has made it so I don't have to be on steroids and antibiotics every other month. I did not start the medication for weight loss purposes, though have lost 10-15lbs, but I attribute that to being off steroids.
Long term? Who knows for sure. Semaglutide has been approved for 9 years. Liraglutide was 2009 and hasn't shown thyroid issues still. Exenatide has been available since 2005.
GLP-1s aren't new and haven't shown widespread negative issues beyond gastrointestinal upset, and that is typically fixed by... drinking more water. The trials in rats have pointed towards potential thyroid cancer risks but those haven't been seen in people.
There is reason to have some healthy skepticism for pharma. But I am sometimes astounded by how unshakably cynical people can be.
I suppose we all could wind up with cancer in 10 years or something hypothetically, due to some unforeseen mechanism. But here's my story.
I went from 200lb to 140lbs at 5' 9" over 3 years. I feel great. I don't need to eat 1000 calories before bed just to sleep through the night. I focus better at work. Blood pressure is back to normal. HDL is back to normal. I can run without joint pain. I feel good about myself. I feel comfortable going on dates. I'm not a near-obese shut-in loner that can't be in a caloric deficit without wrecking my sleep.
Unless there are literally fatal side effects, these drugs seem pretty miraculous.
How can AI absolve us of the need to think and at the same time require special skills to use AI? This seems contradictory.
However AI is similar to cars in this regard. We used to walk, ride a horse or a bicycle to get somewhere. These means of locomotion weren't very fast, but they could take us just about anywhere and they made us very connected to nature. Cars removed all that and forced us to remake our society around cars. We wouldn't be able to ride a horse today even if we wanted to. The only means of transportation is a car on an asphalt road. And yes, riding a car is a separate skill that has nothing in common with walking or riding a bicycle.
I suspect AI will similarly turn our society upside down, so that using your own mind to think won't be possible, just like riding a horse isn't possible today. It's hard to say exactly how this will be achieved, but some tendencies can be seen already: AI produces mountains of slop, making it impossible to read thru by yourself. It's a DDoS on human culture, in essense. If everyone everywhere uses AI to communicate like that, you'll be facing a hard choice: give up your agency, just like everyone else, and use AI to read emails or retreat into the mountains away from this madness.
reply