I need your help. I'm having a hard time wrapping my head around ChatGPT

Don't be afraid of script kiddies. They have always existed and will continue to exist. Just keep pushing and work on the fundamentals - something most prompt engineers do not have a grasp of.

Thank you for the insight!

Certainly, you can email me if you like.

I would second that. Even the best prompt engineers could not do the mental gymnastics to read, comprehend and change the generated code. We are a long way from replacing programmers.

What would be the fundamentals that would stand this AI onslaught?

My advice would be: focus on conceptually difficult but economically useful areas; and try not to spend too much time on the easier stuff if possible (even if the syllabus/grading system values those things). However, if the entire curriculum seems too insipid given new developments, then the opportunity cost would suggest cutting further losses. Hope that helps.

Fun fact: Even Issac Newton was forced through an outdated syllabus; he somehow swam through it, while simultaneously building up the new intellectual framework, which became the new physics. The intellectual climate keeps changing almost every 100 years in some significant way, it always has.

GhatGPT gets lucky a lot. And it sounds confident all the time. So the times it happens to be right get publicized way more. I think eventually, people will learn what it’s really good at, and things will be less hyperbolic.

Some people will learn the system, rest will be enamored for the rest of their lives.

As others mentioned, I believe that as long as there is a prompt to be written, there will be a need for domain experts to write them.

I think our jobs will switch to verifying what GPT-4 and co. are spewing. And yeah, that might get rid of those jobs that are just about mindlessly spewing out HTML. But there are already tools that do that like Wordpress :) Imho this is actually great, we will be able to focus on better, more interesting things, rather than on low level repetitive tasks. So, I think that developers' jobs won't go anywhere, unless you count a developer as a boilerplate factory.

Thank you for your words! It definitely helped me get a better understanding of where I stand.

>I keep reading real, legitimate stories of how ChatGPT was able to all-together replace programmers for projects like developing micro services, websites, etc

Can you (or anyone here) provide evidence to support this statement?

I don’t think programmers will be replaced anytime soon where it should affect your decision to finish your degree. Most things you read or hear about are simple, new code scenarios. In web apps we have things like Rails generators that can pump out boiler plate code, maybe ChatGPTs output is capable of a little more complexity but doing things like that doesn’t happen very frequently and doesn’t save much time. Maintenance of existing applications and their data requires human thought more than teaching a computer to generate net new code.

That's great to hear. I appreciate your insight! For me, hearing it from someone like you who's actually in the trenches makes all the difference.

> but I can't help but feel like my pursuit of a formal bachelor's degree or even continuing to learn data analytics and data engineering will be in vain in light of what feels like major overnight advances in AI.

So, if you think having a degree in engineering will be worthless... what do you think about not having one? Wouldn't that be even worse?

With all due respect, that's a very narrow way of thinking. As I mentioned, I work in Compliance. I already have a degree and career, and simply want to add to that. So, to answer your question, no, not having a degree in engineering would not be worse. It's an option I voluntarily pursued by my own initiative.

Think of a task that a script can do well. Maybe update broken links on a website or a simple CRUD app with some business logic.

Take an evening getting ChatGPT to build it for you.

I was worried about ChatGPT as a blogger until I had it write about things I know about. Now I feel a lot more at ease.

That's good to hear. This definitely helped me understand everything a little bit better.

trying to decide between a data science job and a software engineering one and one of my concerns is the DS job will be much easier to replace with GPT.

If you enjoy data science and are proficient at it you wouldn't have much trouble getting a software engineering job, and vice-versa.

Anecdotally: I do a decent amount of DS work and use GPT to help me move faster through the mundane data tooling work like ingesting new file formats or updating schemas. This enables me to spend more of my time and energy on the interesting science parts (research, algo dev, experiments, etc). I don't see it as threatening the role, rather empowering it.

[usual disclaimers: LLMs hallucinate, never trust their output without review and testing, etc]