Hacker Newsnew | past | comments | ask | show | jobs | submit | AlfredBarnes's commentslogin


It has always felt to me that the LLM chatbots were a surprise to Google, not LLMs, or machine learning in general.


Not true at all. I interacted with Meena[1] while I was there, and the publication was almost three years before the release of ChatGPT. It was an unsettling experience, felt very science fiction.

[1]: https://research.google/blog/towards-a-conversational-agent-...


The surprise was not that they existed: There were chatbots in Google way before ChatGPT. What surprised them was the demand, despite all the problems the chatbots have. The pig problem with LLMs was not that they could do nothing, but how to turn them into products that made good money. Even people in openAI were surprised about what happened.

In many ways, turning tech into products that are useful, good, and don't make life hell is a more interesting issue of our times than the core research itself. We probably want to avoid the valuing capturing platform problem, as otherwise we'll end up seeing governments using ham fisted tools to punish winners in ways that aren't helpful either


The uptake forced the bigger companies to act. With image diffusion models too - no corporate lawyer would let a big company release a product that allowed the customer to create any image...but when stable diffusion et al started to grow like they did...there was a specific price of not acting...and it was high enough to change boardroom decisions


ChatGPT really innovated on making the chat not say racist things that the press could report on. Other efforts before this failed for that reason.


Right. The problem was that people under appreciated ‘alignment’ even before the models were big. And as they get bigger and smarter it becomes more of an issue.


Well, I must say ChatGPT felt much more stable than Meena when I first tried it. But, as you said, it was a few years before ChatGPT was publicly announced :)


It was a surprise to OpenAI too. ChatGPT was essentially a demo app to showcase their API, it was not meant to be a mass consumer product. When you think about it, ChatGPT is a pretty awkward product name, but they had to stick with it.


While I don't disagree, that is unrealistic.


I'd love to hear more ideas for how we can reduce these peoples' power to control our society!


Reduce the power of government if you want to reduce the power certain individuals have over society. Because government is such a single, extremely powerful lever it becomes a singular target of influence and corruption by the rich and influential. Why do you think so many of the rich and powerful move to DC or keep residences there?

The insistence of so many to take away power from Jeff Bezos, who won’t send armed goons to my house if I choose not to buy stuff from Amazon, and giving more power to the government that sent goons to Matt Taibbi’s house the same day he was giving Congressional testimony is an egregious case of missing the plot.


The French have a tool they used quite successfully during the revolution back in the late 1700s.

Jokes aside, unless we go through major societal reforms (that would likely involve a lot of chaos) I don’t see this problem being fixed anytime soon.


It would take one or two not especially complicated law taxing wealth and loans against equity. Congress could do this tomorrow but guess who controls congress. And cap political spending like every other sane democracy.


Support locally owned, small businesses


All you're doing is giving the power to someone else. Giving way more power to politicians like Donald Trump. The reality is, there will always be powerful people, with outsized control and dominion over others; it's baked into the fabric of reality. Thinking otherwise is a utopian fantasy.


well ideally more power should be distributed amongst normal people


Wishing it could be so, flies in the face of everything we know about human nature, power dynamics, and game theory.


I enjoyed this book greatly, I do not enjoy how Stoicism has become the basic meaning of philosophy.

Meditations is also a decent read.


For me this is the running I do. No phone accessible (still with me for safety). Just the repetitive sound of my feet and breathing, lets my mind wander and enjoy the isolation.


100% agree.

A little foil hat conspiracy i supposed, but the big companies saw nobodies become incredibly wealthy over the last decade, and this is the new companies protecting their position by limiting technology.


How do you go about editing via ipad, I'm just getting into this whole world, and am finding it difficult to figure out a system.


I use an app like Terminus or Prompt to access a remote server. I have a keyboard and a mouse for my iPad Pro.


Its cool to see them get it right .....sometimes


This is cool, but feels like a work around that will be patched.


It's just using Cloudflare features for their intended purpose. How is it a workaround? You can even have Gmail configured to send as this email which would be the actually useful thing here because this is just blogspam of the CF feature.


This is not a workaround, this is one of their main product that one should not rely on because Microsoft with their aggressive SPF, DMARC and DNSBL have almost all CF's IP blacklisted. So mail forwarded to your personal MSFT mail account is not guaranteed to arrive and is clearly logged in CF dash.


You're actually right. So, this method might not be considered a long-term solution. The article I linked to is actually part of an online guide where I explain how to set up a company step by step. Readers currently at this stage don't yet have a mailbox (like a Google Workspace subscription with a custom domain). A few steps later, I'll explain how to use Gmail with a custom domain. Then, they won't need email routing. This solution is very useful in this regard.


No, this isn't actually a workaround. This is exactly what email routing does. You can also create a catch-all email.

There are many situations where "forwarding incoming emails to another address" is sufficient. For example, if you don't yet have a mailbox, if you manage several different domains, etc.


Fair, thank you everyone below pointed out I was wrong :)

That's okay I'll try this out in earnest. Thanks for sharing


Wasn't there a story about healthcare companies letting AI determine coverage? I can't remember.


Computers have been making decisions for a while now. As a specific personal example from 2008, I found out that my lender would make home loan offers based on an inscrutable (to me and the banker I was speaking to) heuristic. If the loan was denied by the heuristic, then a human could review the decision, but had strict criteria that they would have to follow. Basically, a computer could “exercise judgement” a make offers that a human could not.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: