» Have there been any mistakes in signature verification for this letter?
» We are aware of two mistakes in our efforts to verify the signatures in the form so far. One person who was not an employee of OpenAI or Google found a bug in our verification system and signed falsely under the name "You guys are letting China Win". This was noticed and fixed in under 10 minutes, and the verification system was improved to prevent mistakes like this from happening again. We also had two people submit twice in a way that our automatic de-duplication didn't catch. We do periodic checks for this. Because of anonymity considerations, all signatures are manually reviewed by one fallible human. We do our best to make sure we catch and correct any mistakes, but we are not perfect and will probably make mistakes. We will log those mistakes here as we find them.
Last night I was blocked from HBOMAX (or whatever brand they go by these days) for being on a VPN. That was the first time I've ever encountered something like that on HBOMAX. I wonder if there is some coordinating event here.
I made several attempts to try to get it to generate something more esoteric. Here is a story about a computer falling in love with a potato chip who becomes a sentient meth addict.
life + death = mortality
life - death = lifestyle
drug + time = occasion
drug - time = narcotic
art + artist + money = creativity
art + artist - money = muse
happiness + politics = contentment
happiness + art = gladness
happiness + money = joy
happiness + love = joy
Having kids is an excellent solution to this feeling.
Besides occupying any time you used to have for unnecessary work, they have an uncanny ability to remind you just how little you actually control.
However you get to the end of the OCD tunnel, the journey is often very worthwhile.
I really enjoyed this. I try to run my team similarly.
Where I disagree slightly is vendors. If the need filled by the vendor is well-defined and low-complexity, sure, I'll go for it. Otherwise, I'm doing it in-house nine times out of ten.
Where this starts to get tricky is when some worthy competitors emerge, utilizing your foundation to scale quicker and more effectively. Then you might wish you had hired more people earlier. But overall, I think starting from this perspective is a lot safer than the opposite.
> It did not reduce student drug use. In face, it backfired and taught kids about interesting drugs that they probably wouldn't have found learned about otherwise.
I will never forget the day in fifth grade when a DARE representative came to our class with a briefcase full of samples of esoteric (to me at least) drugs. The way they were presented made them extremely appealing to me, similar to perusing the choices at a high-end candy store. I don't know for sure if this had any effect on me but I strongly suspect that it did.
The intellectual pursuit of empathy often results in individuals who know what empathy should look like and are good at faking it.
Real empathy requires you to recognize and process your emotions differently. This can be learned over time but usually not by directly attacking the problem.
This is just based on my own experiences.
But I can imagine this is frustrating for you and it seems like you’re already doing the hard work of recognizing there’s an issue. Good luck to you.
I have a bigger problem with how IP law is written and enforced than the actual notion of IP law. It doesn't seem unreasonable to me that IP should enjoy some protections, however, the current enforcements are in many cases downright abusive to consumers and smaller creators alike.
Sample applications are rarely difficult to make. I built a bot as an experiment for my team (https://m.me/PigeonCorps) and we've run into a lot of issues that fall outside of intent classification and response. The most complex issues related to non-trivial chatbots (or atypical chatbots) in my experience have been:
- session management
- horizontal scaling
- dealing with platform changes
- dealing with platform downtime/bugs
- hierarchical data presentation
- matching intents within hierarchical/nested flows
I agree, however, that most chatbots we see are not incredibly complicated, however, I think the typical points that are emphasized in these types of articles (intent classification) are not the hard parts.
Occam's Razor tells me it's because they are being painted as capable of subverting seemingly larger-than-life systems and institutions. I imagine it's an ego boost to a lot of these tech CEOs.