But that's kind of the point - in your culture apparently there is a difference between making unrealistically ambitious claims and being scammy, and in my culture there is not. If someone tries to get investor's money by making unrealistic claims, they may be above the line legally, but in my culture they're clearly behaving morally unacceptably, demonstrating that they are a fraudster, someone who can't be trusted or respected by others in the community, "not one of us", and deserve to be shunned even if they were my friend or relative.
In my mind there is a strong ethical duty to ensure that your claims are realistic, and it's a severe breach of ethics to be irresponsible with unrealistic claims, triply so when others' money is on the line. As far as I understand, in USA being a bullshitter is treated as somewhat acceptable, to me being a bullshitter means you're scum that I'd need to warn others in the community so that they wouldn't work for you or with you. So while in every country there definitely are some "hustler-type" people who do try to initiate startups, they also have issues hiring and local B2B sales if word gets around that they are the type of person who routinely make unrealistically ambitious claims; the world is small, most industries even smaller and people do try to check references of potential partners through friend-of-a-friend-of-a-friend channels - so while a failed startup won't necessarily be held against you, the factors of why and how it failed (and how it affected others) definitely can be.
At least in other places in Europe, and I assume also in UK, it's somewhat common for any sources of financing to new companies (i.e. without a proven steady cash flow), both banks and other investors, to explicitly require personal guarantees from the owner - so while technically you have a limited liability company, the lender will lend you money only if you accept full liability and put your house, car and other assets on the line. As the original post states, "investors spend too much time trying to mitigate downside risk with all sorts of protective provisions". So a company bankruptcy may result in a personal bankruptcy as well.
Hmm, my decades of clicking things lead me to assume a flow from top to bottom, so something with a connector on the bottom is a source that will output data through that connector, and something with a connector on the top is a sink that will accept input data through that connector.
This seems to be begging for a user-switchable "dense mode" (like the fad of "dark mode" switches), which swaps the "buyer-oriented" CSS to the "I-use-this-every-day" oriented CSS.
I think this would have more chances of success if it focused on just one aspect of game creation, instead of trying to solve engine development, gameplay development and content development all at once.
I.e. if an programmer wants to create a game, then getting the artwork in a consistent style is a big problem, and solving that alone (i.e. as a plugin for Unity or Unreal or Godot or whatever) would be both powerful and difficult.
Conversely, if someone without programming skills wants to create a game, then it would be best to focus solely on the AI assistance for gameplay design with a fixed, well-known engine and an integrated well-populated market for content, and assume that they will get the content packs there, leaving that problem out of your scope but making it easy to integrate the AI-generated logic with the pre-made content packs.
Sure, but that's digressing from the key point of the grandparent post about the visual novels - that this median number of "game" revenue is meaningless, because steam sales conflate to entirely different markets i.e. proper games and visual novels, and since the latter outnumber the former but each generally get far less sales (as the market demand is so differently structured) the number isn't informative about the revenue of actual games; and if we want to reason about the game market financials, we have to split the porn in a separate category, as it's too different (money- and sales-wise, it's not about prudeness) to be treated as just one of the genres.
I think there's a consensus that you don't assume that the human experience of "green" is consistent, only that people do have such an experience. We can possibly try to "align" those experiences with communication and referring to a shared real world, but for that an interesting experiment scenario is communication between a person with the common trichromatic sight, a person with a tetrachromatic retina, and someone with partial color blindness, as the experience of "green" for them is not only inconsistent but also likely incompatible, without a possibility to align them.
But organic brains aren't some magic unknowable mush, we know how they work on a low level, we can trace how visual processing is done, etc, etc. As far as we see, the raw computing capabilities of biological neurons that we understand are sufficient to explain the behavior of animals, and as far as we see, biological brains don't do anything that can't be replicated with sufficiently powerful digital computers. So while it's technically possible that consciousness is a specific property of organic brains, we have no evidence at all that it would be the case and some (although not conclusive) evidence that there's nothing special, so unless we identify some difference, those hypotheses are not comparable and we should assume that according to our best current knowledge there aren't any specific properties of organic brains.
There's currently no reason to assume that we're more than a complex biological computer - while it's indeed interesting to explain how subjective experience and qualia arise, it's certainly plausible that this can arise as emergent behavior once a specific type of computer is doing a specific type of computation, and we 'just' need to study that complex computation.
Unless we obtain any evidence whatsoever that this can't be the case, Occam's razor would suggest that this is the hypothesis to explore, without looking for new physics or other unlikely assumptions.
I think there's practical value to knowing what would make a very complex AI system conscious in order to explicitly and intentionally avoid it - we want to create powerful artificial systems to do all the tasks that humans don't want to do, but there's no reason for those systems to have the capability to experience suffering by being 'enslaved' to do all these tasks; arguably consciousness isn't necessary nor useful for any of these tasks[1], and unless it turns out to be an unalienable side-effect of some critical mass of capability, we'd rather not have these systems be conscious.
[1] I can imagine a few roles with relationship-building where expressing consciousness could be useful, but IMHO for all of them faking consciousness would be far preferable than actually having it.
I somewhat disagree, I feel that the prevailing position is that unlike intelligence (e.g. Legg&Hutter definitions) consciousness can not be easily assumed from mere behavior and relies on certain things happening (or not happening) inside the agent.
It is very relevant to keep analyzing and keep trying to get any other answer to this question, because while "appearing like us in every other way, they also experience like us" applies to other humans, as soon as we want to talk about the consciousness (or lack of it) of other actors, this argument can not be applied and we would very much like to get to any other criteria of consciousness which could be applicable to arbitrary non-human agents.
Even if we axiomatically assume that everyone else is not a p-zombie, trying to find any evidence towards your/mine consciousness other than that axiom is helpful as a candidate for such criteria which can be tested and validated.
Knocking the trackers off is a real issue, however, a great non-digital solution to that is the wooden laser-cut boards with physical notches for the cube slots - e.g. https://github.com/cophus/terraforming-mars-board and the many other options if you google or search etsy.
I probably prefer a clear separation of nature and non-nature, with my home/backyard being the latter one, but having actual nature within a walking distance - which requires the non-nature parts to be dense, not sprawling wide.
Worst case complexity matters when the input data can be manipulated by someone malicious, who can then intentionally engineer the degenerate worst case to happen - as we have seen historically in e.g. denial of service attacks exploiting common hash table implementations with bad worst case complexity.
Getting a certain amount (according to their vesting schedule) of stock options, which are worth a substantial amount of money and thus clearly is "good and valuable consideration".
The general criteria for authorship require including the people who worked on the experiments and data for the paper, which can be more important contribution than most of the text in that paper. In other experimental fields, there are papers with dozens or even hundreds of authors, because it can take many people to get to a measurement of a single number in the paper.
I think for most people it's far too late, as there exists at least something on the internet and that something is sufficient - photos can be aged virtually and a single photo is enough, voice doesn't change much and you need only a tiny sample, etc.
And that's the case even if you've never ever posted anything on your social media - it could be family&friends, or employer, or if you're ever been in a public-facing job position that has ever done any community outreach, or ever done a public performance with your music or another hobby, or if you've ever walked past a news crew asking questions to bystanders of some event, or if you've ever participated in some contests or competitions or sports leagues, etc, all of that is generally findable in various archives.
I'd argue that "code that emits SQL" is never an inherent need but a possible development time-saver - we need code that emits SQL in those cases (and only those cases) where it saves a meaningful amount of development time compared to just writing the SQL.
Caveat emptor and your mileage may vary; but unlike nVidia where you could just assume that everything is compatible with everything, for AMD I'd strongly recommend that you try before you buy - consider renting a cloud machine with that GPU to check if the software works for your needs before committing to a large purchase.
We've rooted for that for years, but looking at what AMD does and doesn't do, I've lost hope for this. ĀMD don't seem to want to do what it takes; it's not that they're trying and failing, but they're simply not even committing to attempt to do the same things that nVidia does for their software infrastructure.
Technically we definitely can, but are there sufficiently many people willing to pay a sufficiently high premium for that feature? How much more would you be willing to pay for an otherwise identical card that has the option to expand RAM, and do you expect that a significant portion of buyers would want to pay a non-trivial up-front cost for that possibility?
Yes, some non-linearity is important - not for Turing completeness, but because without it the consecutive layers effectively implement a single linear transformation of the same size and you're just doing useless computation.
However, the "decision point" of the ReLU (and it's everywhere-differentiable friends like leaky ReLU or ELU) provides a sufficient non-linearity - in essence, just as a sigmoid effectively results in a yes/no chooser with some stuff in the middle for training purposes, so does the ReLU "elbow point".
Sigmoid has a problem of 'vanishing gradients' in deep networks, as the sigmoid gradients of 0 - 0.25 in standard backpropagation means that a 'far away' layer will have tiny, useless gradients if there's a hundred sigmoid layers in between.
Why would the license matter in cases where your actions aren't restricted by patent law and you don't need to accept any license and any limitations in it?
If you have a valid patent for e.g "A Device or Appparatus that does FooBar", you get certain exclusive rights for just that - that device is the "patented invention" where making/using/selling one is an infringement of the patent, but the description of that device (for example, the patent text itself) is not. A book about doing FooBar isn't a device that does FooBar, and neither is a tutorial or a blog post - the description of the patent is public, the knowledge in it is public, and you're legally allowed to redistribute that knowledge.
However, using the "patented invention" is an infringement[1], with no exceptions, so if the reader of that book or blog post would like to follow it and actually implement the patented method or follow the patented process in the privacy of their home, that would technically be a violation even if they are unlikely to be sued as the patent holder won't find out about it.
[1] 35 U.S.C. 271 Infringement of patent. (a) Except as otherwise provided in this title, whoever without authority makes, uses, offers to sell, or sells any patented invention, within the United States, or imports into the United States any patented invention during the term of the patent therefor, infringes the patent.
In current neural networks the activation function usually is not sigmoid but something like ReLU ( y = 0 if x<0 else x ), and in any case the computation of activations is not meaningful part of total compute - for non-tiny networks, almost all the effort is spent on the the large matrix multiplications of the large layers before the activation function, and as the network size grows, they become even less relevant, as the amount of activation calculations grows linearly with layer size but the core computation grows superlinearly (n^2.8 perhaps?).
I think there's a reasonable C) which acknowledges that there are multiple jobs where people have an important reason to just have to pee on the spot, and also ridicule gamers for implying that their game is comparable to that.