Live data from Hacker News

National Security Commission on Artificial Intelligence's Final Report

nscai.gov

111–115 of 115 posts

Re: National Security Commission on Artificial Intelligence's Final Report

#111

Earlier quoted context omitted.

Why? What would loosening IP do in this regard?

In its present state the patent/license system does more to gum up innovation than it does to help inventors. This is due to the extreme asymmetry between individual inventors (who have almost no power to enforce IP against big companies) and big companies (who have a virtually unlimited war chest of liberally granted patents that are mainly used to squash competition from smaller businesses.) The options for fixing…

So the solution to a broken system is to break the system in an additional different way?

Re: National Security Commission on Artificial Intelligence's Final Report

#112
post #61

> First, U.S. courts have severely restricted what types of computer-implemented and biotech-related inventions can be protected under U.S. patent law.6 Critical AI and biotech-related inventions have been denied patent protection since 2010.7 Facing uncertainty in obtaining and retaining patent protection, inventors pursue trade secret protection. Trade secrets do not readily promote innovation markets, because trad…

I am wondering, if AI and biotech inventions are denied, why are there so many AI parents given to Google recently?

Hi ilaksh,

Google has a significant patent attorney preparation and prosecution team on its payroll, and they can afford to draft patent applications which have all the enablement and "possession of the invention" text in their patent applications. At the intersection of biotech and AI, this is a significant hurdle to overcome for the solo or bootstrap inventor/innovator.

Applying algorithms to a variety of biotech platforms (genetics, genomics, bioinformatics, single nucleotide polymorphism (SNP), mutation, etc.) is a real challenge for a number of reasons. First, the difference amongst us human beings (a difference in phenotype) is a real challenge for algos to train on, and is dependent upon the underlying datasets or combination of datasets. Second, it's difficult to integrate the algos into existing Hospital Information Systems (HIS).

You can check out these following links for some explanation on this technology convergence and patent prosecution strategy:

https://torreypineslaw.com/convergence-technology-patent-att...

https://www.ipwatchdog.com/2021/01/27/artificial-intelligenc...

https://www.thepharmaletter.com/article/artificial-intellige...

Google has solved some of these problems, whether from its dominant capital position or deep patent prep/pros team, so can successfully obtain granted patents in this technical field.

Re: National Security Commission on Artificial Intelligence's Final Report

#113

Earlier quoted context omitted.

The story you tell about China's rise is exactly the same story I told, I believe. It started with low wages and short-term thinking by rich owners and politicians all too happy to take their bribes, but over time it has turned in a complete inability to compete with China on manufacturing. But one thing is certain: this policy was never a popular demand, American workers have not been happy to let this happen, and A…

I agree with much of what you say. The only issue I have is that you can't ascribe any results (good or bad) to a single variable, or even two, like "wages" and "capitalism". Life just doesn't work like that, nothing does. > my parents' experience in Romania > The actual result was that corrupt politicians sold those factories to corrupt investor One of my neighbors is from Romania. I've had many very interesting con…

The thing about Marxism and socialism as understood by people who've lived through the Soviet Union or the satellite states is that they've been exposed to just as much propaganda about their states being examples of Socialism and Marxism as the West has. It's just as wrong. They were also exposed to propaganda claiming that the state was democratic, but none of them believe that because they understand the word and know how different it was. But Socialism and Marxism had no pre-existing meaning for many people in Eastern Europe (pre-WW2, communism and socialism were prohibited from being openly preached in Romania, for example, under penalty of jail). They had no idea what these terms were supposed to mean, so they did believe that their states were failed socialist states, instead of realizing that not a single socialist ideal had ever been attempted, at least once Lenin took power.

And so yes, to people like my parents the very word brings to mind images of their experience in the 'Socialist Republic of Romania', and of professors and newspapers trying to sell them on how good it was.

They know far less of the history of socialist worker's unions that fought and won for the right to the 40 hour work week or for the right to strike and protest.

Related to examples of successful socialist states, I can't give any. However, rather than debating the reasons for that, I will say that worker's cooperatives are a proven business model all over the world, so I don't see why it's impossible to imagine a successful economy based on co-ops instead of corporation. After all, at one point there were no successful modern democracies or capitalist states in the world, and the very idea seemed preposterous to many lovers of king and country.

Anyway, I want to say that I have greatly enjoyed this thread of conversation!

Re: National Security Commission on Artificial Intelligence's Final Report

#114

Earlier quoted context omitted.

> Why would US salaries start to even out with the rest of the world? On a per capita basis, our GDP is still massively higher than every other large country. The hypothesis is that if most high compensation workers are remote, it opens up those labor markets to more competition from lower cost regions. IOW, as Covid-19 forces organizations to become remote-first, they perforce also become more outsourcing-friendly,…

I don't believe it. People predicted the same thing would happen due to technology enabling outsourcing in the 90s. The only success was at the lower-end of the information economy. Also, while COVID is going to accelerate the shift to remote work in the US, the vast majority will be <100% truly async remote. It's the 1-20% of edge case requirements that are really hard to mitigate. The ability to answer a phone call…

> People predicted the same thing would happen due to technology enabling outsourcing in the 90s. The only success was at the lower-end of the information economy.

I agree, for a variety of reasons, but there is a joker in the deck: will the distribution of required skill change over time? IOW, should we expect demand to grow faster at the low end of the information economy, in the middle, or at the top? I don't think it is too reasonable to expect the distribution to remain fixed over time.

Right now, we've been seeing a lot of growth in demand at the top end, but that was true for the 'service economy' at the start too. The irony of the scenario where increasing automation depresses the wages of the lowest skill jobs because they can most easily be filled with fungible human labor while expanding the demand for workers to fill them due to Jevon's Paradox should not be lost on us.

It is entirely possible that entry-level software work will become as much of a dead-end job as being a customer service representative.

Re: National Security Commission on Artificial Intelligence's Final Report

#115
Most destinations permit you to enter nitty gritty notes about your rivals directly on the screen, where you can get to it with the snap of a mouse. https://qqpokerraja.org/ Numerous destinations will give you point by point insights, for example, how frequently you see the failure, turn and waterway, and how regularly you win when the hand goes to a confrontation. This can be priceless data while dissecting your game to attempt to improve.
Post reply on HN