I was listening to a Lex Fridman podcast where he shared his thoughts on the future of software engineering. He mentioned that engineers will eventually become masters at prompting AI. At this stage in AI development, most people aren’t as efficient as large language models (LLMs), and this gap is likely to widen. In my experience, this seems true: the more precisely I prompt ChatGPT, the better its responses are. Recently, I worked on a bug for nearly two weeks and only managed to fix it once my prompts became very specific. My ability to feed the AI precise prompts improved as I gained a deeper understanding of the bug. I got there by carefully reading error messages and piecing together what was happening. Eventually, I managed to squash the bug with some extra-potent bug spray (code) and submitted a PR. 😄 #SoftwareEngineering #AI #MachineLearning #ChatGPT #LexFridman #TechTips #Debugging #PromptEngineering #FutureOfWork #Programming #TechInnovation #MachineLearningTips #CodingLife
Yeah I 100% agree with this. Using AI definitely speeds up my workflow like 10x
Solutions Engineer II @ Datadog
1moExactly. AI has changed this process a lot. It's not that we rely on "it" to fix things without us understanding what's happening, it's just that AI has become the new Google to help us solve these problems more efficiently. You still have to make sure it works, fact check it, and be sure to understand your code, but the environment, process and atmosphere used is changing.