yeah, I've been trying to get this mindset as well - even though sometimes I feel that I'm "cheating" :)
810 karma · joined March 30, 2018
"DVDs still conquer the market within those sales, and at their peak, DVD sales were over $18 billion, while Blu-Rays made just over $2 billion in a year."
This is the prompt that generates the articles: https://github.com/lfarroco/news-radar/blob/main/src/writer....
- pick HN items with more than 400 votes
- gather their titles in a list
- ask the AI to filter out the ones that are not tech-related (bay-area topics, politics)
- scrape selected articles
- write summaries
- publish static website
other sources are reddit subreddits and rss feeds (official languages' blogs and github's releases page). The AI is quite gullible. That can be avoided by giving it more context and having a review step where you make sure that you are enforcing your editorial rules. Another thing that I've been wondering is having a cheaper model (gpt-turbo-3.5) write articles and then use a more sophisticated to review them (gpt4)
Some things that I've learned so far:
- The AI is too gullible. If I ask it to write a short summary of an article, if the source article is trying to shill a product or service, the AI will replicate the salesman discourse. I tried adjusting the prompt to see if I could make it more critical of content that it is analyzing (by telling it ahead of time that the post might be trying to push a product/service).
- Costs are ludricously low. It costs like 1 cent per 15 articles.
- My next experiment will be with local news. I'm building some feeds with public information from my town (the town's hall official news, the legislators weekly meeting notes, weather reports, waze, etc), and based on that make it generate news items. The thing about it is that its sources will be (nearly) primary - it will not copy content from other journalists (apart from the official prefecture news, which I will need to tell the AI that will be biased towards the current administration). When analyzing the local records, it might be able to catch shady stuff that regular journalists would not notice. Imagine feeding some buying orders from the town and asking the AI ("is something illegal going on here" or "are any of these items overpriced?")
- I see a risk/opportunity for infinite content generation. Example: generate 50 headlines for articles about the Kardashians. The next day, ask for more and provide the last 200 headlines, to make sure that no repetitions occur. It would flood search engines with almost random content. I think that something like that could be useful to fill "holes" in wikipedia, though.
The site that I built is https://dev-radar.com/
Around 2007-08 I read in a(physical) Brazilian web design magazine that you can have two types of design: - for short, ephemeral experience, your design should be striking - the user will just look at the thing once anyway, so you should capture the attention in spite of how ease of navigation. Eg. a website for a rock band tour. - for experiences that will need to be repeated lots of times (eg. using a corporate system, reading the news), you want the design to be boring and predictable.
I think that minimalism has been used to push interfaces that reduce information density - regardless of how the density would be useful in that context!
After the original Material Design video https://www.youtube.com/watch?v=rrT6v5sOwJg the writing was on the wall: if you want to build a google-grade portfolio, create designs that resemble this. After some time users got used to that kind of look and now sites with high density of information look alien.
Layouts with heavy information density (as old.reddit.com) are more user friendly, as you they allow you to just scan the list and be done with it. But are bad for ads.