Reading through Sutskever's deposition is a great example of where the approach to AI mentioned in this post, is failing. The way the world is setup means you need politicians who are hungry for power and willing to do the not so human thing.
Asking for consent to train content, make sure sure AI is "green".. these of idealogical fancies that don't match up to the real world and the winners the folks walking straight over these.
In both cases a smarter model should be help to help offset the greenhouse gas effects and then some when used to discover new materials etc. Hallucinations can be managed, detected and corrected in most cases when doing research and in many cases are just treated as failed attempts as it is easier and cheaper to just ask the model to try again.
The only idea worth chasing here is a community driven and open source approach.
We have this in may formats... my fave is Ben Goertzel's OpenCog .. but there is more. I think Emad Mustaq is doing something similar with decentralised compute/AI.
Again - all bleeding edge is for profit.. not even OpenAI's idealogical foundation was able to be saved from being turned in to a for profit gig.
Honestly.. probably in the top 100 best things China has done. These rings are responsible for kidnapping, trafficking and the deaths of many a victim. Smells like justice to me...and a fair warning!
How to induce mass layoffs without needing to offer severance packages.. without telling employees this is a massive layoff without severance packages being available..
Surprise surprise.. ... not really .. when it comes to Amazon.
It's all about installation.. most folks just can't be bothered to install a different OS then the one their laptop came with. I liken it to the small community of people who tune their cars vs most of the the people who use cars.. just use what they get when they buy.
There is the fact that gaming is a faff with Linux vs something like Microsoft.
It's amazing.. I am a Linux fanboy.. but it's just not a mainstream platform.. and I don't think it will ever be.. and that's OK!
The DeepMind team was essentially forced to publish and release an earlier iteration of AlphaFold after the Rosetta team effectively duplicated their work and published a paper about it in Science. Meanwhile, the Rosetta team just published a similar work about co-folding ligands and proteins in Science a few weeks ago. These are hardly the only teams working in this space - I would expect progress to be very fast in the next few years.
How much has changed- I talked with David Baker at CASP around 2003 and he said at the time, while Rosetta was the best modeller, every time they updated its models with newly determined structures, its predictions got worse :)
It's kind of amazing in retrospect that it was possible to (occasionally) produce very good predictions 20 years ago with at least an order of magnitude smaller training set. I'm very curious whether DeepMind has tried trimming the inputs back to an earlier cutoff point and re-training their models - assuming the same computing technologies were available, how well would their methods have worked a decade or two ago? Was there an inflection point somewhere?
The fact that this was not a thing until recently is mind blowing to me as someone who lives in the U.K.
Doing your taxes is so easy here - takes me 30 mins max, all online and there is even a mobile app to set reminders and give you updates on any rebates or messages.
As someone in the US, there have been numerous free options for filing your taxes for a long time (even officially endorsed ones), and for most people without complex tax situations, it is indeed quite simple and only takes a few minutes. The people who complain about taxes in the US either have uncommon situations, didn't understand the tax implications of an employment arrangement prior to tax season (e.g. became an independent contractor without realizing their responsibilities), or are simply unaware of the free and easy options that have existed for a long time.
Dependent care credit forms, interest/dividend income forms, independent contractor forms, student loan/home mortgage interest deduction forms, medical expense deduction forms.
And that is excluding state tax return, which most have to do too. A sizable group of American tax filers, maybe even most, cannot do their taxes in a few minutes.
I've heard you have to have some sort of registered person do or sign of your tax returns or something ridiculous like this. Something about an entire industry of tax return specialists or something that exist purely for taxes to be done correctly. Of course the US is massive so it might be some random state maybe or something.
Honestly I see America like a collection of large countries as groups of states almost. They are so different in so many ways.
Dystopia vs Utopia - fine we're aware of potential final destinations. Lots of talk on how we get to either. I wanted to take a harsher approach on changes happening in the near term and what they could mean.
Asking for consent to train content, make sure sure AI is "green".. these of idealogical fancies that don't match up to the real world and the winners the folks walking straight over these.
In both cases a smarter model should be help to help offset the greenhouse gas effects and then some when used to discover new materials etc. Hallucinations can be managed, detected and corrected in most cases when doing research and in many cases are just treated as failed attempts as it is easier and cheaper to just ask the model to try again.
The only idea worth chasing here is a community driven and open source approach.
We have this in may formats... my fave is Ben Goertzel's OpenCog .. but there is more. I think Emad Mustaq is doing something similar with decentralised compute/AI.
Again - all bleeding edge is for profit.. not even OpenAI's idealogical foundation was able to be saved from being turned in to a for profit gig.