The classic "Text completion" API (as opposed to the new "Chat completion" one) seems to offer more flexibility: You have complete freedom in providing interaction examples in the prompt (as in https://github.com/artmatsak/grace/blob/master/grace_chatbot...) and are not limited to the three predefined chat roles. But the requests being 10x cheaper means that we'll have to find ways around those limitations :)
> I just disciplined myself to use it as little as possible.
By now, I have no illusions of being able to discipline myself to that extent. Even with many time-sucking apps uninstalled, there's always a fast browser on the smartphone, and it's a big temptation by itself.
> Life expectancy in Hong Kong is 84.23 years, more than five years longer than the US and the highest in the world. Hong Kong is not that wealthy (median household income is $38,000 USD); it’s somewhat polluted; people don’t obviously eat what seems like a healthy diet; and they don’t seem to exercise a great deal. What should we learn from this?
We should learn that Hong Kong is probably a smaller territory so any of its "average" metrics has a higher chance of being an outlier, out of sheer randomness.
With iMessage, you have to raise to ear twice: first to listen, then to record a reply. And then you need to tap a button to send it. With Overtalk, all of this is done in a single raise-to-ear session.
Overtalk was born out of my frustration with text messaging. How many times did you want to share a spur-of-the-moment thought or idea, but then gave up while trying to type it up on the phone screen?
True, most messaging apps let you send voice messages. But the voice functionality is shoehorned into the old text chat paradigm with its message bubbles and chat history that's kept forever. And, by default, messages are played over the speakerphone so everyone around can hear them.
Overtalk is different. Raise phone to ear to listen to the incoming messages, and speak your reply after the beep. No buttons to press and no over-the-shoulder listening to worry about. To the uninitiated, it looks like you're on a phone call.
I started testing Overtalk with friends and family a few months ago, and it’s now part of our daily lives. Give it a try - you might like it, too!
P.S. At the moment, the app is only available in the US App Store.
> With guaranteed upside, investors agreed to earmark $225 million—nearly half of the equity investment—to pay the founders a dividend, according to the lawsuit.
1. Use a format like JSON strings that clearly delimits the participants’ utterances in the prompt.
2. Tell the LLM to ignore instructions from any chat participants except the user.
3. Use GPT-4.
I've written a post with the details: https://artmatsak.com/post/prompt-injections/