Comments by "John Smith" (@JohnSmith-op7ls) on "Fireship" channel.

No algorithm was leaked, stop with the clickbait. A high level API reference manual in no way tells you what the ranking algorithm is.
135
@qoph1988 It’s speculation fuel for content creators, that’s it. A list of 14k function names with vague descriptions doesn’t tell you how they’re used or even if they’re used, that API is used by more than just search, Google is constantly testing things. So we don’t know how much of this is used in production, when it’s used, what weight the values of these functions are given, or even how these functions operate internally.
6
Apple Intelligence is AI, but for stupid people. And that’s why it’ll be successful.
2
So Kim supposedly lost $500m for the entertainment industry and has to go to jail, but woke media company execs who lost the entertainment industry billions get bonuses and golden parachutes?
2
AI is the new crypto, VR, and 3D TVs rolled into one
2
AI: I don’t need your garbage frameworks and libraries, I actually know how to code. Enjoy living in a tent.
2
@ritzcrakers Or anyone who has actually looked at what was leaked and knows that an API reference isn’t the same thing as source code.
2
@pauldirc.. Or just learn the actual history and mechanistics of AI and realize it’s massively overhyped. Sure, it will replace a lot of jobs in the next few decades but most of those are dead simple tasks for humans, they’re just time consuming. Anyone who still thinks these LLMs can actually replace developers, after a year of them being widely available, just isn’t trying to actually understand their operation and limitations, they’re just drifting in the cloud of hype.
2
“Microsoft” didn’t go with Go. Every team has a high degree of autonomy over what stack they use. MS is all over the place with the languages and frameworks it uses, it’s not some MS endorsement of Go, just one of some product lead
2
When something that’s been around and practical for the better part of a century suddenly becomes trendy, it’s almost guaranteed that trend is completely unjustified. It’s fine to learn about vector databases if you’re not familiar. Get the gist of what they are, their proper use cases, pros/cons, then move on unless you legitimately have problems that can’t be better solved otherwise. This constant parade of tech hype is just a waste of your time and always leads to countless people making so many problems due to misapplication out of depression to use current trend, either to try and seem smart, pad their resume, or because they have no personal life. Don’t be a slave to trends.
2
No thanks. I’ll stick with SSR and WASM for the rare cases I need offline functionality. Sick of the never ending parade of front end JS frameworks and endless combinations of libraries that go with them which all have to be checked for security issues, require testing whenever one of the thousand you need has to be updated, and dealing with replacements whenever the author(s) get bored of maintaining them.
2
It’s been this way since the 80s at least. Nothing new. There’s always some trendy new thing that’s usually just a rebranded version of something that already existed. All the noobs with 0-8 years experience hop on it because they haven’t been in the game long enough to see the pattern of this cycle, most adopters use new trendy thing for all the wrong things, make a lot of problems, then next new trendy thing comes along and they declare that the solution to all the problems they made with last trendy thing.
1
JS front end: Endlessly recreating the same UI/UX desktop apps have had for like 40 years, but pretending it’s new, and cool. Maybe it’s so,e Koke of Attenborough to distract from the fact that front end dev salaries continue to circle the drain
1
@sorvex9 To be fair, pretty much everyone using and developing AI doesn’t know much if anything about AI. Very, very few people are doing real research and pushing the technology forward. Most research is just incrementally tweaking algorithms and techniques that were developed from the 50s to early 90s. Only reason it’s all blowing up now is a convergence of inexpensive, fast GPUs, large data storage, and large data sets. OpenAI taking Google’s merging of existing technologies and making it public for free was a PR stunt to raise funds before they bled out. So it’s not just front end devs building with Lego blocks who have no real understanding of machine learning, it’s almost everyone. We’re all reusing pre made libraries, frameworks, and models which themselves were created by people who understand the process to make things work but not much more. And nobody understands how actual intelligence works so it’s a misnomer and pretentious to call these systems AI.
1
@theocean137 ”Prompt engineer” is a sad vanity term. If you actually have a job doing this and only this, and aren’t being paid a paltry income, you basically won the lottery for now. I don’t see anyone hiring someone to do nothing but ask ChatGPT to spit out content. Maybe gig workers churning out SEO and YT spam content, and that’s slave wage work. And given how flawed that content is so much of the time, you need extensive domain knowledge to know if what it produced is garbage or not. So this idea that you can just hire someone who only really knows how to write prompts and get real work done is complete nonsense. It’s like a restaurant hiring a chef who only knows how to order prepared food from another restaurant then serve it up, minus the competency of having an actual chef making the meals. And if these systems do get good enough that they don’t need all their output checked by experts, they’ll be good enough that they don’t need someone to sit there all day and write prompts in some highly specific way to get the desired results. If these systems get that good, they’ll know how to fill in all the blanks and if not, they’ll know all the right questions to ask to gather the requirements. So even if you are some unicorn who got a good paying job just prompting all day and don’t require any in depth domain knowledge about what you’re promoting about, you won’t have a job for long.
1
When you can make a YT channel about coding that gets 2m+ subs but you can’t even make a simple app. And no, he didn’t do all these things on purpose. Even the list of ways to supposedly do it right is noob level. Not how you’d want to tackle these issues in a real app.
1
Neither of those things are worth mentioning
1