I would question the narrative that humans lack the capacity to verify the output and would instead argue the people lack the incentive to verify the output.
The response of many mathematicians to the recent dump is a good example: verifying these proofs amounts to unpaid labour for OpenAI and wastes time that could be spent doing publishable work which ultimately results in money or personal success. The slop factor also compounds the work required to verify the output considerably.
For mathematicians, programmers or anyone, if the work required to deal with slop passes the limit, it is no longer in their own self interest to use LLMs. The expectation that people will use LLMs for the betterment of humanity against their own financial interest is baffling.
Humans are not immortal and cannot spend all their time into review (especially unpaid). Even today, there’s so much knowledge around that you have to be specialist of a narrow domain to get to the frontier. Even in computing which is just approaching a century of existence.
On the contrary, AI has sucked all the air out of the room. There are still a hundred thousand normal consumer apps waiting to be built, but nobody's building (or funding) them anymore, they're all off chasing this new thing.
This reminds me of the early 1980s when synthesizer music became popular. People back then were saying things along the lines of "anyone can make music now, you just press a button and you have a no. 1 hit". But the reality is that, just having a synthesizer doesn't make you musician but a synthesizer in the hands of a musician can make a big difference.
They're referring to supply chain attacks. Taking over open source libraries through social engineering and adding hidden malicious code. Becoming increasingly common.
One could disable updates of dependencies - especially near-real-time updates. Just updating a week after everybody else will have a lot of potential harm with libraries that have a large user base.
Most of the dev work out there in the broader corporate world (not startups) doesn't care about "shipping fast". Writing code is not the bottleneck.
When your day consists of lots of meetings with humans who have technically vague requirements, nobody ever meets you halfway. Your primary job becomes bridging those knowledge gaps mostly by yourself. You have to buckle down and make independent decisions that neither the stakeholders, nor the LLMs, can help you with. You get paid because the organization trusts you with lots of information that only lives inside people's heads.
In that situation, the LLMs are only really useful for familiarizing yourself with the existing code. They are completely irrelevant to writing code. You'll have entire two-week sprints to make very precise changes to only about a dozen lines of code, but the coding is not even 10% of what you're paid to do with that time.
In my opinion, this is what senior software engineering always was. Everyone raving about LLMs are basically code monkeys working in sweatshops.
Not actively looking but always open to explore opportunities. Prefer startups,
greenfield projects and true innovation. Generalist with deep experience that learns and moves quickly.
Performance and efficiency have been neglected as everyone threw every available GPU and trillions of dollars trying (and failing) to create AGI. Though the results have been impressive.
But good enough for pennies in an instant is very useful.
It had a charm the full game was missing. It reminded me of a puppet show and how large something so low fidelity can feel when filled out by imagination.
Most of us have worked our whole lives to be writers in some domain: Code, documentation, specifications, product requirements, corporate communication, etc.
If we don't want to be writers, then we have to be editors. And editing is an entirely different job and it's not an easy one. In many ways it's harder.
Especially when LLMs love writing novels when all we need is a short story or less.
For decades, velocity, time to market and features all drove the industry. But AI puts speed in everyone's hands.
But quality is still hard. Maybe harder than ever.
reply