I would be curious to know if the WGS was done using short read or long read. There is a lot of hope that long read is "just" what is needed to catch these type of cases nowadays (not sure about this particular one).
I'm in the process of getting my first batch of long reads, but I am skeptical that this is the "just" what's needed. There is little doubt that long read > short read, but I think that computational techniques for both need to be improved significantly.
There is already some clinical evidence to support my hypothesis. The first clinical long read trial at Kansas City Mercy showed a 10% bump in diagnostic rate, which is great but not fully solving the problem: https://news.childrensmercy.org/unlocking-answers-faster-chi...
The contrast in my opinion comes more from the fact that 50 years prior to the space age people rode horses as a standard mean of transportation (more or less). It’s underwhelming to not see the line going from horse to rocket continue on the same path 50 years later.
Wouldn't these markets be the perfect "crime crowdsourcing" platform? Let's say a bad actor wants XYZ to happen so they organise a bet on XYZ not happening. Then make sure as many different accounts that they control directly or indirectly bet a lot of money in the same way. Now anyone willing to win the prize can decide to make sure it does happen and get the money (at least a high multiplicator of what they bet themselves).
Am I missing something? Maybe it's already being used in that way... in that case, it would be quite scary.
In recent months I changed views and shifted from the desilluioned "this is a casino" mindset that is described in this article to a "we need this now" one.
An example in this article [1], the US can now unilateraly decide to prevent an individual anywhere in the world from having a functioning financial life and this because of the quasi (western) duopoly that is Visa and Mastercard. Nothing against the US in general, this is simply too much power to put in a single decisional entity, whatever/wherever it is. The "crypto" related systems now seem like a needed extra option to the current payment system (the same way cash is almost always an alternative to credit/debit card payment and vice versa)
If, after all these years, Crypto has not become a stable store of value that can replace a fiat currency and let you buy your groceries with it, why do we suddenly 'need it' to do that now, and what would change that will enable that to happen?
It's still not possible because of the high speculative aspect (even though stable coins offer some form of alternative but since they are anchored on the dollar, not so much). Change will happen when more people are impacted by unilateral decisions. If all of a sudden the US decided you/your business/your political party/your government is not aligned with their views, they could make sure that Visa/Mastercard/Paypal/Stripe/... stop accepting payments for that given group. Then you realize there are very little alternatives ready to go in that case (for online payments there are, but for physical ones not so much). So far this hasn't seem like a possibility at all, but, in my view at least, this has changed this past year.
Understanding crypto from this type of international context focused on these sorts of issues is where it indisputably makes much sense and is seeing indisputable adoption. Low and slow but end of the day to a very large and growing problem, bitcoin+ adoption or a mass civics readjustment in the US are the solutions. Which is more likely?
So it’s an inefficient tech with a mess of problems and uneven adoption but if you want to send $1-$1mm anywhere in the globe you can. That’s very powerful tech and the implications are about as important as anything else from cryptography hitting public adoption. And all of those have been consequential.. see 30 year fight about e2ee.
This is very important point, that people from US and EU oversee.
I live in EU for many years, but due to my birth country being sanctioned I can't use any financial instruments like investing or even simplest savings deposits. Getting mortgage or loan is also much more harder for me, even tho I have much better financial situation than average person in that country. Apart of that I need occasionally go to the bank in person to proof the bank that i'm good person with valid documents under the threat of freezing my funds and closing my accounts.
Funniest thing in that is all these sanctions are issued by EU and US, and not by the country I live, where i'm pretty welcomed.
That's just the money cycle. Ray Dalio explains it in his new book, "How Countries Go Broke, Principles for Navigating the Big Debt Cycle, Where We Are Headed, and What We Should Do" (https://www.principles.com/)
The short explanation is that every major currency goes through a cycle; and what you're seeing is the late stage in a currency.
Speaking of TF-IDF I once added it “after” the recommendations to downscore items that were too popular and tended to be recommended too much/with too many other items (think Beatles/iphone) and inversely for more niche items. It might be too costly too do depending on how you generate the recommendations though.
Not sure if you’re joking but a relatively small datacenter I’m familiar with has reduced oxygen in it to prevent fires. If you were to break in unannounced you would faint or maybe worse (?).
Not quite - while you can reduce oxygen levels, they have to be kept within 4pp so at worst, will make you light headed. Many athletes train at the same levels though so it’s easy to overcome.
That'd make for a decent heist comedy - a bunch of former professional athletes get hired to break in to a low-oxygen data center, but the plan goes wrong and they have to use their sports skills in improbable ways to pull it off.
I think this whole em dash topic should lead to some deeper (though not very deep) conversations:
* If it was not widely used before where/how did (chat)GPT picked it up?
* If it was widely used, then it shouldn't be a topic at all. But, there seems to be informal agreement that it wasn’t widely used.
* Or, could GPT have inferred that even though it's not widely used, it's the better way to go (to use it). Which then makes one wonder about the whole probability of next token idea. Maybe this line of thinking falls too short of what might be really going on internally.
* If it had picked up something that is widely used but in the wrong way, it should make us pause (again) about the future feedback loops these LLMs, which aren't going away, are already creating. Not just in terms of grammar and spelling but also in terms of way of thinking and seeing the world.
The training sets of most LLMs contain a copious amount of content from Libgen (or now: Anna's Archive), where em dashes are frequently used in literary writing.
It's used a lot in formal writing (academic papers, books etc) which are probably a large portion of chatGPTs training. If the HRL was done by professional writers then it was probably additionally biased toward using them.
People are more casual on the web. It's sort of like how people can often tell when it's me in IM without my name because I properly use periods while that's unusual in that medium. ChatGPT is so correct it feels robotic.
It’s the most likely explanation I believe. I have no idea about the content distribution of the training data but I would have assumed twitter and Reddit content would completely dwarf the literary content. Somewhat good that if it’s indeed not the case!
It isn't about wide use. It is about a character that almost no-one enters explicitly. Nearly all usages are copy paste, or inadvertent/unintended conversion by an application such as Microsoft Word that converts regular quotes to smart quotes, etc. In that respect, we see that an AI is performing identically to a real human. An AI does not and most likely would not add see a purpose an em or en dash to any text, unless it was an article about em or en dashes, or they knew the person they were speaking with uses en or em dashes.
Maybe do something close to what I like to believe the brain does and have a meta model wrap a "base" model. The meta model gets the output data from the base model (edit: plus the original input) as input plus some meta parameters (for example the probability each token had when it was chosen and/or better which "neurons" were activated during the whole output sequence which would include the Persona they mention). It's then the meta model that generates new output data based on this input and this is the output that is shown to the user.
Can you describe the "meta" model more ? afaict it seems like you are describing a "router"? I think what you are thinking of is essentially what MoE does, or in diffusion, a sort of controlnet-like grounding (different exact mechanism, similar spirit).
I've said this somewhere else, but we have the perfect test for AGI in the form of any open world game. Give the instructions to the AGI that it should finish the game and how to control it. Give the frames as input and wait. When I think of the latest Zelda games and especially how the Shrine chanllenges are desgined they especially feel like the perfect environement for an AGI test.
And if someone makes a machine that does all that and another person says
"That's not really AGI because xyz"
What then? The difficulty in coming up with a test for AGI is coming up with something that people will accept a passing grade as AGI.
In many respects I feel like all of the claims that models don't really understand or have internal representation or whatever tend to lean on nebulous or circular definitions of the properties in question. Trying to pin the arguments down usually end up with dualism and/or religion.
Doing what Chollet has done is infinitely better, if a person can easily do something and a model cannot then there is clearly something significant missing
It doesn't matter what the property is or what it is called. Such tests might even help us see what those properties are.
Anyone who wants to claim the fundamental inability of these models should be able to provide a task that it is clearly possible to tell when it has been succeeded, and to show that humans can do it (if that's the bar we are claiming can't be met). If they are right, then no future model should be able to solve that class of problems.
Given your premise (which I agree with) I think the issue in general comes from the lack of a good, broadly accepted definition of what AGI is.
My initial comment originates from the fact that in my internal definition, an AGI would have a de facto understanding of the physics of "our world". Or better, could infer them by trial and error. But, indeed, it doesn't have to be the case. (The other advantage of the Zelda games is that they introduce new abilities that don't exist in our world, and for which most children -I've seen- understand the mechanisms and how they could be applied to solve a problem quite naturaly even they've never had that ability before).
I'd say the issue is the lack of a good, broadly accepted definition of what I is. We all know "smart" when we see it, but actually defining it in a rigorous way is tough.
When people catalogue the deficiencies in AI systems, they often (at least implicitly) forgive all of our own such limitations. When someone points to something that an AI system clearly doesn't understand, they say that proves it isn't AGI. But if you point at any random human, who fails at the very same task, you wouldn't say they lack "HGI", even if they're too personally limited to ever be taught the skill.
All of which, is to say, I don't think pointing at a limitation of an AI system, really proves it lacks AGI. It's a more slippery definition, than that.
> It doesn't matter what the property is or what it is called. Such tests might even help us see what those properties are.
This is a very good point and somewhat novel to me in its explicitness.
There's no reason to think that we already have the concepts and terminology to point out the gaps between the current state and human-level intelligence and beyond. It's incredibly naive to think we have armchair-generated already those concepts by pure self-reflection and philosophizing. This is obvious in fields like physics. Experiments were necessary to even come up with the basic concepts of electromagnetism or relativity or quantum mechanics.
I think the reason is that pure philosophizing is still more prestigious than getting down in the weeds and dirt and doing limited-scope well-defined experiments on concrete things. So people feel smart by wielding poorly defined concepts like "understanding" or "reasoning" or "thinking", contrasting it with "mere pattern matching", a bit like the stalemate that philosophy as a field often hits, as opposed to the more pragmatic approach in the sciences, where empirical contact with reality allows more consensus and clarity without getting caught up in mere semantics.
> The difficulty in coming up with a test for AGI is coming up with something that people will accept a passing grade as AGI.
The difficulty with intelligence is we don't even know what it is in the first place (in a psychology sense, we don't even have a reliable model of anything that corresponds to what humans point at and call intelligence; IQ and g are really poor substitutes).
Add into that Goodhart's Law (essentially, propose a test as a metric for something, and people will optimize for the test rather than what the test is trying to measure), and it's really no surprise that there's no test for AGI.
reply