Bypass with archive link

Paywall bypass

All I'm saying is there's events that lead up to the point of singularity, and then we are beyond it. As we've discussed it more I've cared less about the semantics haha.

Kurzweil's view of the singularity will absolutely be wrong. The process he came up with for the prediction time is solid, but he's wrong on so many levels about his expected outcome and the reality when the time comes. But his timeframes are pretty good.

He also has consistently described bldck holes incorrectly over the years, so I don't think he really went too far in depth over why people before him used the word singularity.

But either way tech is moving at an exciting pace.

There was a deleted tweet from a venture capitalist that was posted by someone (reddit post was removed by mods here). Some of the more advanced information seemed to be factually incorrect, but it also claimed Google's Sundar Pichai called "code red" over chatGPt and most likely the future with GPT4. The New York Times (archive link for paywall), among others, published an article like the next day about the code red.

What it appears to mean is Google is in a somewhat critical situation of do-or-die when it comes to search. Monetizing an AI will be difficult, and search ads are the core of their revenue. They are all hands on deck now, channeling brain power to this situation.

Now back to that tweet, within the same point it also said Microsoft will incorporate gpt into Bing by next year. If that's true, then yes not only is SEO going to non-exist or be vastly different, but it is also happening extremely soon. However keep in mind, while that tweet had the code red call correct, it also claimed GPT4 will be AGI and Turing complete. So it could have just been someone that heard about the code red a day earlier than it was announced, and then fabricated the tweet to extend legitimacy to the rest of the points in it.

I'm my opinion, judging by the fact the VC mentioned earlier did in fact tweet this and delete it, I'm inclined to partially believe some of the more general information in it. But it sounds like a CEO hyping their own chat bot (Sam Altman) but I think Microsoft will probably incorporate GPT4 into Bing somehow, and Google will be forced to do the same to stay relevant.

I don't have physically have access to Kurzweil's book, but here's a YouTube link [7:39] where he agrees it's a point where everything is different after than it was before. An epoch is usually a distinctive period of time following an event that sets it off. The singularity being the "event" and post-singularity being the era. We often call it "The Singularity" but you can also refer to it as "a singularity" which further shows it is a single event, not a span of time. There will certainly be many many events that lead to that moment.

When I say event, I'm under the assumption we will probably retroactively note the approximate time (or exact time who knows, future tech could be crazy) that it happened. When either one extremely significant transformative technology makes and impact that in hindsight we couldn't predict; or several technologies converge over time and at some moment in time humanity is changed. We can only predict that time now, but it's impossible to know until it has already passed and things are different.

​

Honestly at this point though let's just agree to disagree. I don't see any possible convincing arguments that haven't already been made from either side.

You could use other terms, such as Transformative AI. It describes the exact situation you're expressing. I don't want to sound like a nitpicking idiot or anything, but it's an important distinction that the singularity is in fact a moment and that we're either pre-singularity or post-singularity. You can make the argument that we're already post singularity, I'd probably disagree, but the opinion is your own.

I was just clarifying because it pops up in this sub often that people have this idea of the singularity and to be honest I'm not sure where that idea is coming from other than maybe being a feedback loop within this sub and similar online discussions that began as a misinterpretation of why we use the word singularity for a specific use-case.

Of course you're free to ignore me altogether haha, to each their own.

If you can (or care to), I'd be happy to see some references stating it isn't a moment in time but a period of time.

Thanks, I try to not be too wordy in comments which can make me sound like much more of an asshole than I intend to come across as. It's just a definition that has been skewered, and while the distinction isn't a hug difference, it's important so we don't get people claiming we're "in the singularity" right now. You're either pre-singularity, or post-singularity. There's no "in" and it's probably not going to be as of significant "event" as several things preceding it, and many many things following it.

No, the technological singularity is by definition a moment in time. You don't live in the singularity as a timeframe. It is an instant moment when it happens, a single point in time that passes as quickly as it happens. Again, it's a hypothetical point in time, meaning you can live pre and post singularity, not within. Not within even briefly. It's a single dimensional moment. The post singularity events could very well be eventful (I imagine it will be) but the actual snap of the finger when the singularity passes will not be anything monumental at that moment.

We are not even close to our institutions and humanity as a whole not being able to keep up with technological innovation at the moment either. I'm an optimist, and I can't wait. But unless we have sudden AGI next year and it takes control of our progress and future, we aren't "in" a singularity. I believe we are in the elbow of the curve, and things are about to take off. But I don't believe we've passed the moment of singularity because my life still feels very much predictable.

>AI is not even a necessary component.

For one, we are talking directly relating to AI. Even without AI, it means a technology that is so transforming that we haven't yet anticipated its impact (something like femtotech?). That could also arguably be some kind of medical break through that changes our entire perspective on life, say total immortality or something. Doesn't matter it's irrelevant to the discussion.

Second, the only definition is in direct comparison to the term used in physics, by which you aren't "inside" of a singularity the moment you cross the even horizon. I'm not trying to be overly direct or rude here, but you can't just use examples from physics to describe this and expect it to make sense when you've misused the terms.

From your original comment:

>My personal definition of a technological singularity is a point in time when we're experiencing so many black swan events that the future becomes impossible to predict at shorter and shorter timescales

Your thought process behind increasing occurrences of black swan events is perfectly acceptable as passing the event horizon. I like that reference, I've used it before. But crossing an event horizon does not equal being inside of a singularity. The technological singularity is a blip in time, not something you sit around in and chill for a while like we currently are in the "space between singularity and event horizon."

Anyway, that's about enough from me on the subject. I hope I didn't come off as rude or anything.

>Are we already in the midst of a singularity?

No. That violates the definition of a singularity.

When/if we reach that moment, it will have already been behind us in an instant. You don't experience it, it is just a hypothetical way of describing our (all collective humanity) ability to predict what is to come. We may never reach a moment of singularity in the future, while at the same time having already surpassed that moment by today's standards. It will not be an event, at least not by the actual definition of the technological singularity.

Some timeline where an ASI instantly synthesizes the entire universe (unlikely but whatever) is not "the moment of singularity" but rather just some bizarre outcome that would have nothing to do with what we mean by the term technological singularity.

The singularity is not an event. I understand romanticizing it, it's fun to think about. But it's just not what it is, the term has a very real and concrete definition. Though it is hypothetical, it's still a real term that has a widely accepted definition.

>I said we've crossed the event horizon, which puts us 'inside' the singularity.

That is essentially the same thing I claimed you said. The event horizon is normal times, you would unknowingly cross that barrier. In a physical sense, that would mean time slowing to an observer watching. I agree we are likely past that barrier/threshold in that more technological break throughs happen in shorter and shorter timeframes and eventually (the moment of singularity) there is a hypothetical infinite amount of technology being created AKA impossible for us to comprehend right now. But being within the bounds of the event horizon does not mean being inside of a singularity.

>I didn't come up with it, Kurzweil did as far as I know.

He didn't invent the comparison to physics, but that's besides the point. His definition is exactly what I stated. And I was referencing your comment directly, where you said you have your own personal definition...

>There's no 'actual' definition. It's a hypothetical/speculative.

There quite literally is an exact definition, and it isn't speculation. I'm not sure where you're getting that from, but it's a term that is widely used but this sub misuses it continually. It is a hypothetical thing, but not a speculative definition.

>My personal definition of a technological singularity

​

>The event horizon.

I mean, you can have your own personal definition if you want, but that makes no sense. Not trying to sound rude or anything. An event horizon is not the same thing as a singularity. That's not having your own definition, that's just calling one thing another thing for no reason, specifically because we have definitions for both of those things already.

I agree with the comparison of being at or beyond an "event horizon" in terms of AI. But the singularity is an infinitely brief measure of time in which the moment we reach it, we have passed it. That moment, by actual definition, is when AI reaches a greater intelligence than all collective human intelligence. It probably won't even be a significantly impactful moment, it will just be a reference. We look at it now as some grand moment, but it isn't. It is just a moment where it's impossible for humanity to predict anything beyond it because an intelligence greater than all of ours exists, so we can't comprehend the outcome at this time.

The individual capacity of a human to not be able to predict what comes tomorrow has no bearing on whether or not the singularity has passed. Even if all human beings trying to predict what will come tomorrow are wrong, that still is not the singularity. It's a hypothetical time in the future that based on on today, right now, we know 100% we cannot make a prediction beyond because it's mentally impossible as a direct result of our brains being incapable.

It's interesting to consider that we may never reach the moment of a technological singularity either. If we merge with technology and increase our own intelligence, we could forever be moving the singularity "goal posts" similar to how an observer sees someone falling toward a black hole forever suspended, yet the subject falling felt a normal passage of time from event horizon to singularity. We may forever be suspended racing toward the singularity, yet at the same time having reached and surpassed it.

>For me, the real question is what exactly is GPT-4 going to be able to do.

Yep, this is the burning question. Since it will have the most SOTA scaling methods/tuning, if it's a much larger scale model than GPT3, we will may be able to extrapolate a lot better on where we're headed with subsequent models in the future. We currently don't really have any baseline to compare outside of GPT2 to 3 which isn't really that useful to compare anymore.

Found it. It's a good resource I think, someone had posted it on reddit but I don't remember where that thread is. I had to search my browser history for the link.

I still haven't been able to convince any friends to try it out, they see they have to sign up for it and login in and decide it isn't worth the effort.

People won't see this but it's correct. While I didn't go as far as saying my post is the most important thing in this sub or anything dumb, you have to make it enticing to read otherwise it's a wasted effort typing any post if it just gets buried.

Definitely underestimating how little the general public cares about it

Well define slowly? By quickly do you mean 2023 we will have AGI and capitalism will be gone? Because I have my doubts there. I think there will be a period time we adjust to Transformative AI over the next few years. And years is a long time for people to suffer unemployment.

To be clear, I don't think prompting will be important in the future, even near term future. It's familiarizing yourself and getting used to using AI to gather answers. Like how some people still don't use the internet to its full potential, a lot of people will just use AI without pushing the limits of what they can accomplish with it. As in don't be afraid to start learning new things using AI.

I see how I may have not made it clear that learning prompting probably won't be the future. It's more learning what AI can do now, so you can have a head start on getting comfortable with what it can do later.

It's basically just a post saying go use AI, only I made it 1000x longer.

Haha that's a perfect real-time example of the clueless nature of some people. Which, to each their own I guess.

Behave? Haha ok. I won't bother engaging here.

Like explain what the singularity is? Or explain my post?

I saw a website that has a whole huge list of them, I'll see if I can find it when I'm back at my computer later today.