Hacker News Reader

Comment by dang | Hacker News Reader

Discussed at the time:

Generating code from natural language is closer than you think - https://news.ycombinator.com/item?id=1912530 - Nov 2010 (80 comments)

View on HN

From the top comment:

> I teach computer science and have a particular fondness for introductory CS. The reason Stephen Wolfram is wrong, wrong, wrong about this is that people that have never been taught programming can't express themselves precisely enough in their native language, either; and even among those of us that have been programming for decades, when we express ourselves in natural language we can be very precise but it takes a lot more work and becomes a lot more unwieldy than just writing out our instructions in [pseudo]code.

I used to think the same thing, but I what changed my mind was the noticable increase in precision that came with ChatGPT 4. Before I felt like a monkey - not I actually get things done they way "I intended".

So you can go to chatgpt and say," I want to make a new search engine that is better then my competitors in performance, resource consumption, and cost. It must be the best in class for relevant searches, using a new hyper efficient search algorithm and data structures. It must have a pay to use API so I can sell ad space. It needs to have a marketable name with best selling branding. I also would like this search engine to have email, an online multiuser Microsoft word clone with the fastest live editing feature. This platform must be written in terraform, html, go lang, and node js using trusted dependencies and be completely secure using best practices. I also need end to end tests for the deployment of this product, dashboards to monitor it's uptime, performance, revenue, etc in AWS. My initial budget is 100k. Given the technical realities of my request also provide a summary for any limitations that arose, and technical specifications of this product."

Dang dude, why aren't you a billionaire competing with Google, bing, etc tomorrow?

Talk of moving the goal post.

People are having a discussion about the ability of LLM to generate correct code from imprecise natural language descriptions and how good GPT4 was at that and your rebuke is that it can't do cutting edge algorithm research and an unnecessary ad hominem.

No ad hominem if you reread both your post and mine. But okay let's shift the goal post back, remove the need for new algorithms and ask chat gpt to make you the code to compete with Google. Might not be a billionaire but still could be a millionaire.

From the second comment: *I would be fascinated to see several hundred years down the road how natural languages and computer languages have comingled and evolved into something new. I'd be inclined to believe that bringing natural language to computers won't just be a one-way street.

Order update: Your wish is arriving hundreds of years early.

Reply on news.ycombinator.com