As someone who's been "vibe-coding" successfully at work I can say that what made me a believer in AI tools is agentic instructions. It is the difference between not being able to accomplish much and actually getting stuff done. There have been several articles published on this topic on HN. I would really recommend implementing that in your repositories.
I was in the market looking for a used SUV for my wife to haul around our kids last year. Dealing with dealers who wouldn't budge even a few hundred off the inflated listing prices, the interest rates on financing and then the insane insurance costs was not pleasant. I have an old Subaru beater that I was also thinking about upgrading but after this whole fiasco I decided to spend some money to fix body and small mechanical issues myself and drive this thing another 10 years hopefully. It's just not worth it.
Google is damned-if-they-do, damned-if-they-don’t. One side screams that Google Search is dead because Google doesn't innovate. The other - Google AI innovation is killing the web.
Yeah it was incredible in the beginning because it was so novel. Now it's just annoying. Half of the dialogue is repeated and it takes forever to get a point across. Never used NLM, but I wonder if that's something that can be tuned out?
Yes, all the time. I use ChatGPT to help me get a general sense of direction when I'm learning something new. Once I have a few potential paths to explore, I combine regular Google searches with ChatGPT prompts to make sure my understanding aligns with reality.
My parents have had Google Fiber since 2011 or 2012. It's been a joy. The uptime has been great and the price has pretty much remained the same for 1Gig. In Los Angeles I pay $90/mo for 100mbps Spectrum connection which drops once a week and has gone up from $40 to $90/mo in just under 2 years. Worst of all is Spectrum has complete monopoly and I have nowhere else to go.
I got my first programming job in 2008, seemingly right before the wave of hiring freezes. I dodged the bullet back then, but I wish I could compare that time to the past two years. That said, it feels very rough out there. Just last week, one of my LinkedIn connections shared a Venmo link, asking for help to pay bills after being laid off, struggling to find another job, and running out of savings. Be thankful if you have a job.
Every technological advancement that we've seen in software engineering - be it in things like Postgres, Kubernetes and Cloud Infrastructure - came out from truly novel ideas. AI seems to generate outputs that appear novel but are they really? It's capable of synthesizing and combining vast amounts of information in creative ways but it's deriving everything from existing patterns found within its training data. Truly novel ideas require thinking outside the box. It's combination of cognitive, emotional and environmental factors which go beyond pattern recognition. How close are we to achieving this? Everyone seems to be shaking in their boots because we might lose our job safety in tech, but I don't see any intelligence here.
Maybe I'm missing something vital, but how does anything that we've seen AI do up until this point or explained in this experiment even hint at AGI? Can any of these models ideate? Can they come up with technologies and tools? No and it's unlikely they will any time soon. However, they can make engineers infinitely more productive.
Yep. Grew up in the 90s Russia. Every spring when the snow melted you had bodies show up in nearby forests. It was such a common thing people started calling them snowdrops (after the first flower of spring). Everyone tried to keep cash in USD and you often got better rates from the guys selling currency in the streets. These guys were dying like flies. I've lost count how many times we'd find dismembered hand or foot while at a family outing in nature. Nobody reported any of this stuff because the cops were often bought by these criminal organizations. It was truly a wild west.
In my experience -- I'm a staff engineer -- it's basically a senior engineer in a technical leadership role, but not a manager. I spend 50% of my time working on code and providing technical direction for seniors. The rest of the time is spent in architecture and planning meetings. I don't do 1-on-1s. I don't do personnel reviews. My manager handles all of that.
Polymarket is a new thing and it's throwing a lot of political pundits off their tracks because they treat it as political sentiment indicator, but it's all just gambling and manipulation.
It's not interesting, it's preposterous to think that you can make that money and just kick your feet up every day and twiddle your thumbs. Yeah, there is going to be fucking stress. That's why the pay is so high. My non-FAANG job is 100% constant stress day-in and day-out and I don't make even half of these comps.
Maybe if you stopped with the sarcasm and thought about this critically you'd see that the solution isn't taxing everyone to death, but reducing the out-of-control government spending. Because at the rate the government is spending money, you will be paying this "capital gains" tax in 2 decades on a much smaller balance.
You know what else was a "great idea" that was supposed to be only for the wealthy class? The income tax [1]. 100 years later and all of a sudden everyone is paying it.
I was implying that AI can be useful in software engineering because hallucinations are expected and actively addressed by engineers. It's less of an issue in comparison to automating strategic business decisions by what is essentially a "black box" and not knowing if it hallucinated data in order to arrive at its decision.