>“Anonymization is hard, and you’ve got to have the right technical experts at the table to come up with the solutions,” Adkins told Wired.
So Google, with hundreds if not thousands of security engineers, basically infinite budget, more than 500 published papers on security and privacy, multiple security teams, including one (Project Zero) that has discovered at least 2,000 vulnerabilities in the last 10 years, and with actual experience in privacy applied to its many apps and solutions... is unable to anonymize data when required to avoid a lock-in?
Well at least Anduril refers to a sword used by the “good guys” so, industrialization aside, it is a somewhat reasonable name for a US defense company.
Palantir really doesn’t make sense. It’s like developing a bulletproof vest and naming it Horcrux… sure, in the HP world a Horcrux does protect you from death, but it is very associated to evil, so no one who is really fan of the saga would choose that name, especially for a publicly known product.
Then again maybe they just asked an LLM “I have a surveillance solution. I want to name it after a fantasy concept. Any ideas?”
Also, a gravity-like model makes even more sense in ancient times which were really constrained by distances, unlike modern economies for which this is more true for goods than services (although there are gravity effects for services, too). In a 100% goods economy, with information itself traveling at the speed of a man or a horse, everything becomes gravity-like.
Maybe it's just me but this post is written in a confusing way.
I think the argument here is:
- laws are slow
- innovation is faster
- if you let laws regulate AI you're basically playing catch up
- therefore, let technology regulate AI in real time: once a decision is made, it is immediately valid.
From an efficiency standpoint this may make sense, but the important question is: who makes the decisions that are then implemented in software?
AI platforms are generally provided by private operators, so there are two actors here: the State and the AI provider.
Assume the State decide to completely deregulate AI.
Let's take deepfakes.
As long as an AI provider makes money out of deepfakes, and they do not fear sanctions (remember, there is no government regulation), why would they ban them? Reputation risks? Well, there are many AI providers, so as long as they can produce deepfakes that cannot be traced to a specific AI platform, why bother?
It's one thing to say that laws are slow, quite another to say that democratic accountability and general concerns other than making money should be thrown out of the window in favor of fast but obscure processes.
Yes, markets have existed for a lot of time, well before established economic theory.
I don't think this surprise anyone, especially given that Adam Smith, who is known as the father of modern economics, wrote a book describing stuff that was obviously already happening before his ideas were published. Not to mention Italian bankers, tulips bubbles, gold coins and so on.
However, this doesn't mean that things do not (or cannot evolve). Markets are observed by someone, a theory is born, markets evolve, maybe following a theory, an existing theory is generalized, and so on.
You lose the concept of DNS forwarding. Usually, if your company has example.com, your DNS server is authoritative for example.com, which means it will actually contain (fqdn,ip) entries belonging to example.com, and it will forward requests for other domains to other DNS servers, possibly one DNS server per domain.
If you remove DNS servers from the equation, you need to write down records for other domains, too. This means you have to chase every domain for changes in CDN configuration, hosting provider or ISP migrations, IPv4 to v6 migrations and so on.
You don't have PTR records, which means you can't find out a name from its IP address.
You also miss other features of DNS, like SRV, MX and so on.
More subtly, you lose the ability to control DNS resolution over systems you can't control. If a DNS server says host.example.com is 192.168.0.4, a Windows desktop, a Linux server and your toaster will agree on that (especially if no local cache is enabled, but even then TTLs apply). If for some reason you cannot control a particular machine, you will never get it to consider that new DNS record. This can happen for a lot of reasons.
Disclaimer: I haven't been able to read the full article because I don't have a subscription.
Yes, there are alternatives to GDP, but context is key.
If you just to make more meaningful charts or rankings of top 10 countries, you just switch to another indicator and that's it. This could also impact, to some extent, real economic choices.
Other applications are more difficult.
For example, debt substainability. Governments issue bonds to finance stuff, but they eventually have to repay at least interests. This introduces a link between GDP and debt growth. If you replace GDP with another index things get more vague: a country may have the maximum happiness, but if it issues bonds and it cannot repay its interests, it's going to have some trouble anyway.
Monetary policy is also difficult to decouple from GDP. Alternative indicators often mix many aspects, which means the central bank can either
1) target the whole index, which means it could try to influence not just interest rates, but education, healthcare and so on (it essentially blends with the government), or
2) keep targeting inflation, but through non-GDP estimates, which means the price "signal" is mixed with other "noise".
If the only person who can get the money is you (or your partner or children or whatever), it’s fine as a form of compensation for potential damages.
If anyone, including your surgeon, can take that life insurance policy based on your life, things can go bad pretty quickly (hint: what happens if a profit-maximizing surgeon would earn a lot more money from your policy than from his regular job?).
Engineering is not just about the tests. It is also about following standards, making decisions based on data and requirements not just personal preferences or trends, and especially being aware of the legal and ethical implications of what you're building.
That's why there is so much emphasis on signing or stamping projects in the "ordinary" engineering world: you're literally telling a potential third party (which may include a court of law at some point) that you reviewed the work and can be held responsible.
This doesn't rule out team work of course (just think about civil engineering, it's not something you can do alone most of the times), and this doesn't mean there are no unethical or even corrupt engineers, but there is in most cases a single person who is responsible from a professional, ethical, and legal point of view.
In my experience this never occurs with programming unless you're dealing with SWE in engineering-adjacent sectors (aerospace, ...), and even then I'm not sure as I have no personal experience in those areas.
>Moreover, ownership of data cannot be left solely in private hands but must be appropriately regulated. Data is the product of many contributors and should not be treated as something to be sold off or entrusted to a select few. It is necessary to think creatively in order to manage data as a common or shared good, in a spirit of participation, as Saint John Paul II already suggested regarding collective goods. [128]
It does mention IP concerns, but that's not the greatest existential threat posed by AI.
You can but it’s not part of a plaintext standard. You could certainly build a text editor which can decode a base64 string, but to transform that into a picture you need something implementing the GIF (or JPEG or PNG) specification, otherwise it’s just a binary blob.
I think the real limit of plain text is pretty obvious: you cannot embed pictures in it.
It’s like SMS vs MMS or modern chat. With pure text, you can at best add a link to a picture (which could get rotten or inaccessible for other reasons), but you cannot directly graphical content.
So Google, with hundreds if not thousands of security engineers, basically infinite budget, more than 500 published papers on security and privacy, multiple security teams, including one (Project Zero) that has discovered at least 2,000 vulnerabilities in the last 10 years, and with actual experience in privacy applied to its many apps and solutions... is unable to anonymize data when required to avoid a lock-in?