If these tools get to a point where they redefine software development (they unequivocally have not redefined software lifecycles at all companies I have witnessed) then they will build up dependencies for these tools at the enterprise level while slowly ratcheting up the prices for all tiers. Finally they'll reduce alternative development options and features, requiring access to these tools to compete at a high level.
I don't think they'll ever be so good that they beat a truly great software engineer. But they'll get into the 60th or whatever percentile where they become an honest player in the off shoring conversation.
Before all this happens software engineer salaries and soft power will tank, purely off the rhetoric used as a shield so the executives can avoid the guillotine.
Remember we absolutely cannot unionize because then we'll earn less money. No matter what we must not unionize.
What's the reason behind this? If agents can code, doesn't that effectively take away your job as an engineer and even your work? I mean, if I can use your agent to build myself your company, how are you going to survive in the future?
1. They think they can control this and sell you a service that can make any app you want nearly instantly.
2. They plan to use any profit they make from the service or sale of the company and invest it in stocks/bonds/real estate after and retire before they themselves get replaced.
3. They plan to use this to keep the team small. One talking point of AI is that it could allow far fewer people to make a valuable company.
4. They don't really know what will happen in that world.
That is the question I keep asking myself. Many makes theories about getting rid of software developers. İf this can be achieved, then software companies wont exist. Every new day, there should be a new OS like windows, a new word, excel etc. So, it is not engineer shooting themself from the foot, but software companies does ?
The cost of using their agent to replicate the same agent may be astronomical. Maybe the Reflection folks figured out that will bring them enough $$$ to retire afterwards.
Funny how HNers are now concerned but were all 'meh' when Silicon Valley has been trying to automate away my job for years. The world is a safer place when you take fallible humans out of the loop, don't ya know?
Launching Reflection. Our team pioneered major advances in RL and LLMs, including AlphaGo, AlphaZero, and Gemini. We're building superintelligent autonomous systems. Starting with autonomous coding.
We believe that solving autonomous coding will enable superintelligence more broadly.
We are about to find out that if you let a human read every book, then only let her play games and code programs, what is going to happen. Would she have general intelligence just like everyone else?
I thought it was interesting that neither the headline ("Reflection: A Path to Superintelligence") or the subhead ("Reflection is an AI company building superintelligent autonomous systems") bother to mention coding. After all, if you're not on the "superintelligence" hype train, what are you even doing?