Ask HN: Is there anything worth developing with AI? | Hacker News Reader

Ask HN: Is there anything worth developing with AI?

2 pointsmf2hd6 comments

It seems to me that humans already developed everything that makes money, multiple times.

Apps? The market is full, hundreds of apps are doing the same. Another hundreds are country specific (and doing the same). Bike rental clones, Uber clones, whatever clones, if the business model works it's already cloned.

Games? 12000 games released on Steam this year so far. No one cares. GTA6 is the game for 2026.

Is there anything out there we cannot do (and already done) without AI?

Is there a killer AI app?

I'm talking about coding specifically, because it seems like this is the easiest thing to do with AI. The training data is excellent (all the training data code is tested and working, thanks humans). The output of the AI can be automatically tested for validity (working or not). I don't think anyone going to write code from now on (for what?).

Corporations can develop new features quickly for existing products. Great, but is there anything missing? From fb, gmail, whatever? Are these software so complicated it's impossible to maintain anymore by humans? The tech debt is so big a new feature takes months to develop? Or what? Even if this is the case, and the AI can do everything it in a day, what is the killer feature?

It will be useful for testing security, we need AI to protect against the AI.

And it's going to replace a lot of programmers, especially at big companies. The lazy programmers automate everything, but this time it went too far :)

Next year someone will make millions on a 2D game.

The year after that someone will sell a search company for millions.

The year after that the most popular game in the world will be playable in a browser.

There will be new ways to get around town, around the world, find a place to stay, find babysitters, buy groceries, there will be new computers, gaming consoles, clothing brands, gigs to make money in ways you never thought like holding keys for people or taking photos of things at certain times, there will be new apps that help people get paid instantly instead of slowly through banks etc.

Plenty of room for innovation everywhere you look.

The evidence is how bad and clunky most experiences are and how much is entirely missing in terms of products, services, and alternatives.

Every category leader is subject to disruption - I can’t think of any flawless product or service.

I am not a programmer or software engineer by any means, but I use AI as a high powered assistant who needs robust oversight. Helping me craft FOIA requests and processing datasets returned to continue digging and researching, reverse engineering, de compiling, etc. It is, in my use cases, a moat killer (ability to avoid moats of non open source software) or multi tool/"swiss army knife".

To your point, corps are constantly chasing more dollars and TAM, whereas I can build what I need for me, I control it, and I can freely share the output with others for them to build off of.

https://craigmod.com/essays/robot_blood/ speaks more eloquently on this than me.

Yes, it is very useful for these kind of things. I hacked my Zyxel router's firmware with it, and reverse engineered the Amiga Kickstart rom, things like this, it's fun :)

But where is the money? Would you start a company based on this? I wouldn't. I just solve problems with it I'm too lazy to do otherwise.

I get paid to solve problems. If it helps me get paid to solve problems, it has economic value. I will pay for a tool to help me solve problems, based on the cost and what economic gain I can realize in total. If it helps me solve my own problems that are not business or profitable problems, I'll still use it, just differently (ie "cost efficiently").

Don't Call Yourself A Programmer, And Other Career Advice - https://www.kalzumeus.com/2011/10/28/dont-call-yourself-a-pr... - October 28th, 2011

Threads: https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...

it’s joyless, but AI seems great at old legacy software integration and sustainment. Companies have old packages they cannot move off very easily/quickly, lots of legacy data, and AI is very good at coming up with creative ways to sustain/integrate them, and in the process automate the most skull-numbing/laborious/high-risk aspects of that work. (I’m dealing with +20yo packages right now, doing things I NEVER would have agreed to even attempt in a pre-AI world.) Proprietary data formats, obscure protocols, nothing is a show-stopper anymore. No fun, but pays the bills.

You said it: AI defense against AI, since it is already so pervasive.

Especially so in jobs. AI resume reviews scanning AI-generated resumes.

For years, I wrote down ideas for new and liberating tech applications [0] and services, but found that I was only counteracting the abuse of users by sociopaths in control of large-scale technology.

Each advance is usually snapped up first by military, for more efficient killing, or scoundrels, for more efficient scams, along with some non zero-sum benefits, and it takes time and a lot of effort to undo the damage, though killing is irreversible.

Is there anything we can do?

YES, it's called meeting real needs that aren't fast buck or planetary scale like the "big tech" giants. They go unaddressed, and doesn't a "power to the average person" technology like we dreamt of in the past and never showed, or was dumped, make that possible?

When the first wave of "globalization" came and we all asked "Where are the Jobs" (2004) [1], I thought that FAB would enable people to create bespoke products that met customers' unique needs. Oh well.

Hear me well.

There are many urgent and unmet needs, and we are once again facing a solution in search of a problem, or worse, redefining problems and creating new ones to satisfy the solution.

[0] https://kk.org/thetechnium/what-technology/

What Technology Wants, by Kevin Kelly

[1] https://pastebin.com/raw/WwPErQ2U

Reply on news.ycombinator.com