Even as a self-motivated learner I fail to see the bigger impact of AI. For a “virtual Feynman” I would prefer the online video courses and books which exist without AI. The best I expect an AI to do is to answer my questions and confirm my understandings. At AI’s current state I can use it as a better search engine but due to hallucinations I can’t expect reliable answers yet.
Asking it “3.11和3.8哪个大” (meaning “Which one is larger, 3.11 or 3.8?” in Chinese) and it answers 3.11 more than half of the time. I assume it’s because Python 3.11 is larger than Python 3.8. While it does work in its native language English, this failure doesn’t give me much confidence in its reliability, as we don’t know why it works in one language but not the other yet.
My bank app asks for different tokens for different operations. A code for login, a code for transfers (the code needs to be generated with the payee account number as input). So it’s not a problem of tokens vs SMS.
Every VPN advertised in China takes DPI in account and there is a very huge community around how to circumvent GFW. The previously popular protocol was Shadowsocks (until GFW blocked fully random traffic) and now I use Hysteria2 (which masquerades itself as QUIC) and it has not been blocked yet.
In case your favorite VPN provider does not provide these protocols you can buy a VPS and deploy one of these protocols by yourself.
Is there any sort of criminal risk to playing cat and mouse with GFW censorship? Or are you generally free to just do your best to subvert the limitations put in place?
A quote from someone I know working in a related agency for “cybercrime” in China:
> If you just set it up and use it yourself, at most you will just be given a verbal warning and urged to rectify and confiscate the tools. If you set it up and give it to your friends to use, there may be risks, but in reality, law enforcement agencies usually don't have extra time to manage these small things. If you set it up for your friends to use, the traffic of three or five people is very small, just normal access to the external network, and there is no other illegal behavior such as money laundering, telecommunications fraud, etc. No one cares. If you are the airport owner, that is, the VPN commercial service provider, then your legal risk will be very high, you will be involved in illegal operation of telecommunications services, because this requires you to apply for a license, and VPN commercial service providers usually earn large illegal income, may be sentenced and confiscated for several years. Note that the sentence is due to its illegal operation. Once he has the qualifications, then he will be a legal communication service.
> Because such behavior can be subject to administrative penalties at most in the law, which is what I just said, warnings or orders to make corrections and confiscate tools, and at most fines, but generally not. The fine will be determined based on your attitude of admitting your mistakes. If your attitude is good, there will basically be no fine. In addition, if there is illegal income to be confiscated, such as illegal income obtained by browsing the Internet and conducting cross-border money laundering or cross-border telecommunications fraud or computer network crimes with international illegal organizations, it will be confiscated.
Funnily enough, he says almost everyone used VPN-like technologies, even within the regulatory agencies.
Is the arbitrage opportunity worth it? It doesn't seem like it is considering the limiting factor being the court system, and you can't really arbitrage off cheaply off miniscule enough violations to profit more than a few one off cases. Damages must also be present, so maybe a service to service offering where one party disparages you online and then you take Meta to court for not suspending that party would work, but thats fraud, and if found out you risk quite a bit for sums that seem to not really exceed more than $15,000 across the US (from the article).
I personally find such comparisons unfair. A good comparison should optimize for each device configuration, which means use a model within the VRAM limit and quantize to 8 bits where it boosts performance etc and avoid shortcomings of both devices unless necessary.
I’m surprised it would corrupt the database. The database could have included a unique identifier as well as in the journal and refuse opening the database if the identifiers don’t match.
Uniquely identifying a database and its journal/wal files would prevent just one of the five mispairing cases they mention. And it's the least likely case.
LLMs does have that, or at least it’s very likely that we will eventually be able to manipulate LLMs in a modular way (see https://news.ycombinator.com/item?id=40429540). One point remains: humans learn language with much fewer tokens than LLMs need, which suggests presence of a priori knowledge about the world. The LLM metaphor is finetuning, so babies are born with a base model and then finetuned with environment data, but it’s still within LLM scope.
You don’t want to be responsible for some random receipt 10 years ago that allegedly commits tax fraud, allow influencers to manipulate you into buying things or supporting campaigns that you immediately regret 1 minute later, receive lots of spams in your email and social accounts, etc, and the stakes get higher as you climb up the social ladder. Many of these don’t matter for people with little stakes as they have nothing to lose.
I mean I respect him for his contributions but what about being more specific with the contributions, like “I brought the idea” or “I supervised the research” or even “I managed the lab and arranged for the equipments with which my students conducted research”, any would be perfectly valid contributions to science. But it seems like the academic norm is just to boast “I was an author of these shiny new papers” and the details of how are obscured away almost intentionally.
