4 ms·
I think this equally applies to all knowledge workers. LLMs are not going to cut the human out of the loop just yet, but building them into the workflow is goin
by dougmwne 4y ago
I think this equally applies to all knowledge workers. LLMs are not going to cut the human out of the loop just yet, but building them into the workflow is going to end up being so powerful that anyone not using an LLM will be uncompetitive.
I see this being at least a few years out. Performance is stil scaling up and cost is still scaling down. It’s going to require collecting a lot more specialized training data to produce fine tuned model. Business will need to adapt and people will need to learn how to accelerate themselves. But eventually, I see this going the way of the introduction of the computer to the office. No productive person will work without it.
As a thought exercise, I’m going to assume you’ve used same GPT model to produce some code before and been impressed in a narrow kind of way. Then project that forward a few years, through GPT-4 and GPT-5. Add some better training data and a better understanding of how to fit the LLM output into day-to-day work. Maybe add the ability to fine-tune on the existing code base and design documents. The trend-line is at least going to hit junior dev level, so the job will probably turn into a senior dev supervising a “team” of LLMs.
- username3 4y agoPeople will still use Excel.
- jstummbillig 4y agoNothing will ever beat the horse.
- analog31 4y agoYes, Excel supercharged by LLM trained on all the world's spreadsheets.
- noduerme 4y agoCan't wait to short the first bank to try that one.
- dougmwne 4y agoI think this is a great point. Why do people use excel? I think it’s because they take not old obsolete software, but actually one of the best ways for non-technical people to create simple applications. You can basically turn out a custom inventory tracking system in a few hours for a small business using this ubiquitous software that almost everyone has some skill in. I think that same accessibility is going to be a major driver for these natural language models. I’ve seen great examples where people can basically “program “ in English by having a conversation back and forth with the LLM and refining its output of code.
- noduerme 4y agoI'm personally fine with that as long as integrating a "junior dev level" AI coder doesn't create more debugging for me. The problem I see right now, as this creeps from writing narrowly defined functions to spewing out whole architectures is that it's turning a small footgun into a giant one that gets much, much harder to debug and understand. If you don't even know parts of your own program, how are you supposed to know why something is going wrong? Ask the AI? And then ask the AI to fix it? The danger here is that people start generating code they can't even read or debug; it may work in most cases but when it fails, it will fail spectacularly.
- PeterisP 4y agoI think the "junior dev" analogy is spot on - just as many junior devs are talented and productive enough to be really dangerous without some supervision to ensure that the result fits in a reasonable architecture that won't result in burnout-driven suicides for its maintainers, you have to apply the same level of senior guidance and supervision to LLM-generated code. Of course, there will be companies who'll get the one bright-eyed young person whom they can exploit the most, and get them to apply ChatGPT to build all of their systems, cheerfully reimplementing all the most popular architecture flaws and (e.g.) security issues that are seen in the many repositories of old code used in training the model.