After trying out several different versions, I’ve crafted this prompt that’s proven effective for summarizing news from various outlets:
Please follow the instructions below step by step:
1) Go through the Reuters website and collect the top headlines. DO NOT SHOW ME YET. Proceed to the 2) task.
2) Go through the Forbes website and collect the top headlines.DO NOT SHOW ME YET. Proceed to the 3) task.
3) Go through the Fox News website and collect the top headlines. DO NOT SHOW ME YET. Just save it to the last task. Proceed to the 4) task.
4) Go through the Yahoo News website and collect the top headlines. DO NOT SHOW ME YET. Just save it to the last task. Proceed to the 5) task.
5) Head over to Bing News and gather the headlines for stories. Make sure to exclude any headlines from Reuters, Forbes, Fox News, and Yahoo News since we’ve already covered those. Now proceed to the last task.
Last task) With all the information collected, identify and analyze the three most popular topics across each category (Business/Finance, Technology/Science, Politics, and World News Events) considering all the sources you’ve reviewed. Summarize these topics and prepare a newsletter to brief me on your findings.
Very cool. Thanks for sharing. If you have it in a thread on your GPT page, do you prompt the next day (and all after) “Do the same as above, but try to be thoughtful about repeating news items I’m already aware of.” ?
I can see a kind of arms race happening between advertisers trying to stop bots from pulling news off sites that use ADs, and app developers who’re using AI to extract the valuable pieces of information automatically, so that no human needs to experience to the ADs.
I think the advertisers are going to loose, unless/until they adapt some sort of strategy like “product placement” (in movies, etc) whereby they’re able to camouflage their ADs as content so cleverly the AI can’t tell the difference. And when they do that everyone is going to hate it and hate advertisers for their dishonesty.
So there will be a lot of innovation done in making AIs capable of “detecting” when something is an AD or not. Most likely companies like OpenAI will end up declaring “AD Detection” to be an “unsafe” technology and ban it.
In fact, many changes will come. GPTNews, for example, cites sources, but if the entire article can be read by the AI, it will be possible to talk directly to the AI about the article without needing to read it.
Perhaps the best approach for news sites is to expose some of the content (a few paragraphs beyond the headline) to AI bots, so that users need to access the site to see the full content (along with ads).
Currently I think it’s legal for me to go read some news sites, and then paraphrase all that content in my own words, and put it on my own site, without any ADs, or use my own ADs to make money.
I don’t think the law can ever really stop this from happening, because no one can really prove whether I used AI to do this or not.
I’m pretty sure it’s illegal to do a sentence-by-sentence paraphrase and call it “my own” (copywrite), but if 100 websites are consumed by AI and then the information is put into auto-generated bullet-points for smoother consumption, I just don’t see how that can ever be made illegal.