If anything, I would say that that's a very optimistic take. The hype train is strong, but that's largely what it is once you look at the details. What we have right now is impressive, but no one has shown anything close to a possible path from where we are right now to AGI. The things we can do right now are fancy, but they're fancy in the same way good autocomplete is fancy. To me, it feels like a local maxima, but it's very unclear whether the specific set of approaches we're exploring right now can lead to something more.
I'm not convinced, and neither is Sam Altman himself [0]. Also, if that projection holds, and that's a big if, the purported breakthrough would cost 10^6 times as much as GPT-4 took to train. That's over 100 million dollars [1] times a million. That adds up to over 100 trillion dollars, in the ballpark of four times the GDP of the whole of United States.
The thing is that it looks like, or perhaps I should say it's "understood" at this point, that transformer's abilities scale pretty much linearly with compute (there is also some evidence they scale exponentially with parameter count, but just some evidence).
Right now there is insane amounts of money being thrown at AI because progress is matching projections. There doesn't seem to be a leveling off or diminishing returns taking place. And that's just compute, we could probably freeze compute and still make insane progress just because optimizations have so much momentum right now too.
I think that's part of the carefully-crafted hype messaging. Close enough to get excited about, but far enough away that by the time we get there people will have forgotten we were supposed to have it by then.
Yeah, that's my number one question, too. Sure, he happened to be appointed the manager of the team who cracked intuitive algorithms through deep learning, but what does he know about superintelligence? IMO that's a completely separate question, and "foundation models continue to improve" is absolutely not related to whether or not an intelligence explosion is guaranteed or not. I'd trust someone like Yudkowsky way more on this, or really anyone who has engaged with academic literature on the subjects of intentionality, receptive vs. spontaneous reasoning, or really any academic literature of any kind...
Does anyone know if he's published thoughts on any serious lit? So far I've just seen him play the "I know stuff you don't because I get to see behind the scenes" card over and over, which seems a little dubious at this point. I was convinced they would announce AGI in December 2023, so I'm far from a hater! It just seems clear that they're/he's guessing at this point, rather than reporting or reasoning.
Really he assumes two huge breakthroughs, both of which I find plausible but far from guaranteed:
With nearly-limitless intelligence and abundant energy
I’m not that much crazy older than you, but I do remember well being a teenager at the peak of the dotcom craze, being a heavy internet user, and even back then having a vague feeling that the ad riddled, desperately monetized (and shitty) websites built on the back of mostly investment capital run amok felt a little shitty as an end user. It’s hazy now and lost to time (the internet is not actually forever it’s like ~10 years old at most now) but i vividly remember pages that became harder and harder to navigate because of invasive ads, and sites that’d somehow embed malware on your computer that’d spam you with weird porn popups when your parents used the machine - it all feels vaguely similar to now, albeit much sleeker. Adware in my opinion is bordering on malware to the point I find the definitions indistinguishable. It collapsed then for good reason, and IMHO similar conditions as to now. I don’t want to live through that as a fully grown adult with a tech career now, and it worries me a lot.
What arose from the ashes of that bubble event became great so maybe a reset is needed, but for me personally, it’d be a disaster.
Wikipedia is from 2001, Archive from 1996. IRC still exists as do retro games or the demo scene. There's still some of that good stuff around, but you can't really imagine it being founded today, at least not with the same cultural enthusiasm. Such an optimistic time...
I've been thinking about this too, perhaps it's my age (43) and the fact I vividly remember earlier times.
If I were to pick an inflection point, a point at which the internet started going to shit, i'd say it was around 07/08 with the birth of the iPhone and Appstore. That's when "pay to publish" really started to take off.
>
If I were to pick an inflection point, a point at which the internet started going to shit, i'd say it was around 07/08 with the birth of the iPhone and Appstore. That's when "pay to publish" really started to take off.
That's very plausible.
I additionally want to add that before the iPhone, having a locked-down device where the vendor decides which app(lication)s you are allowed to install caused huge outcries and shitstorms.
Example: Microsoft's initiatives for "Next-Generation Secure Computing Base" (formerly Palladium) [1] and attempting to enforce a TPM on computers (keyword: trusted computing).
When the iPhone came out, this all suddenly became perfectly accepted.
I don’t know if Palm was technically a walled garden, I’m guessing you could load from wherever, but practically it was and I don’t think anyone had a problem with it. Not sure if it’s a counterpoint but something to consider.
> I don’t know if Palm was technically a walled garden, I’m guessing you could load from wherever, but practically it was
According to this Reddit thread [1], you could easily install applications to a Palm from a memory stick. Additionally, I am not aware that Palm applications needed to be signed by the device producer (i.e. the device producer could not decide which applications are allowed vs forbidden on the device).
The Palm was wide open, there was zero DRM, I had a Palm IIIx in high school and spent hours browsing the web for fun freeware/shareware to HotSync onto it over serial. I also spent my time waiting at the bus stop writing my own bus time tables app directly on the Palm using yBasic
Palm even provided the source code of the built-in apps like the calendar etc as sample code with their SDK, leading to a huge ecosystem of freeware or shareware "Calendar Plus"-type applications that just added small quality of life features
Palm was an open platform. You could install apps from wherever you like, or write your own.
I learned 68k assembler in my teens for the purpose of cracking Palm apps for my Palm IIIe. I obviously had no credit card, and I had very little money.
For me it was around this time, but 2013 when the Snowden revelations came out. The internet became creepy because of all the spying and collection. It was a distinct change in my attitude I remember distinctly.
Not sure how it was 2007, but tumblr feels like relief nowadays because it still has some glimpses of that old internet. You can customize a tumblr down to the HTML, including JavaScript.
Not using the social parts (likes) etc. though. I'm using it as a way to share photos and what I'm up to with family without forced login. Like a homepage basically. It is not trying to distract visitors with pointing to other blogs either.
This is interesting to me because I vividly remember the Summer of 2006 as being one of the most fun times I had on the Internet. I wasn't aware of these details but probably would've also said around 2007 or 2008 there was a vibe shift.
One of the possible names for Gen Z that was thrown around about a decade ago was "iGen", a reference to the iPhone and noticing there was a cultural shift in those that came of age on/after 2008.
I'm in my very late 20s and I do miss this, too. I used MSN more than ICQ, however. I still use IRC just like I did when I was ~13 years old, without much knowledge of English.
> I so badly wanted that future of the Internet, but somehow we ended up in a place where corporations ate it all.
The high bandwidths we got today are exactly the result of a full commercialization of the Internet. If there was no money to earn here, we would still be stuck with dial-up connections and had no YouTube and Netflix.
I understand that advertisements and online tracking suck but it’s the result of consumers not willing to pay a penny for many online services. But I think that’s already changing with all those subscription models and SaaS businesses out there.
Blaming consumers for the state of things and for “not spending money” is a common refrain, and honestly is gaslighting and revisionist. The internet functioned fine for over a decade, profitably, without invasive adware like we see now, which is to the point of total degradation of the core service itself - this is not the cause of consumers, but rather the current mindset of the wall street landscape.
As far as “not willing to pay for it” what do you call subscription models of popular language models like chatGPT? I would pay an embarrassing amount for a google that worked like google did 10 years ago. That isn’t my fault that product doesn’t exist anymore, the demand is there.
This angry tone isn’t directed at you, I just find it so frustrating that people believe it’s such a binary choice. Google had the literal monopoly on tech talent and knowledge for 20 years and decided to divert that into the most cannibalistic and predatory business model around. Can you imagine had they directed the same efforts to making an actual competitor to AWS? I am speaking as a career cloud infra guy and business owner running on cloud - as much as I hate MS products I’d sooner migrate to azure than ever spend a penny on a google cloud product for anything more critical than running the office coffee maker. That, I think, was a tremendously bad decision for the internet, and the decision absolutely was not binary. Make a good product people find useful and people pay for it. That’s how the market has worked for all of human history, there’s nothing different about the internet.
> The internet functioned fine for over a decade, profitably, without invasive adware like we see now
What time period are you even talking about?
In the mid nineties the Internet was largely a research network between university computers and paid for by tax payers. The internet only started growing exponentially with commercial services appearing in the late nineties. This was the time when people started to demand high speed Internet connections (ASDL) and were willing to pay for that. But this was not the case for its services. Google's primary business model way back in 2000 was already showing ads related to your search terms. Even back in 2003 Google introduced the free GMAIL service that showed ads based on your email content. That's 21 years ago.
IMHO it is your view on the Internet history that is being "revisionist" and "gaslighting". Take this from sb who had his private Internet access as early as 1998.
> In the mid nineties the Internet was largely a research network between university computers and paid for by tax payers.
Your timeline is off by at least half a decade, and things were changing very rapidly in that time. By the mid-'90s, NSFNET was formally dead, after years of accepting commercial traffic. Local ISPs for home users started popping up and AOL opened its access to USENET in 1993.
The push for residential broadband also started almost immediately; @Home was offering residential cable internet in 1996, it was the future at the time, just not very evenly distributed. This was well before PCs had the processing power to do standard-definition video.
> By the mid-'90s, NSFNET was formally dead, after years of accepting commercial traffic.
Mind you: Your statement says nothing about profitability. Until 1995 the NSFNET was the most important backbone of the American internet and it was paid for by the US government. Of course, other countries lagged way behind the US' development where the Internet still remained primarily a research network until the late nineties.
The German Telekom e.g. was still pitching its BTX service in 1995 (kinda like teletext for modems) that could be used for shopping, online banking and stuff -- stuff which you could not really do on the WWW yet. I remember my first experiments in browsing the WWW happend over a BTX gateway and it was awfully slow even by the standards back then. The first DSL service only started in 1999/2000 in Germany in the form of a test program where you were required to answer market research question to get a cheap, subsidized service.
(I mean we are talking about the _World Wide_ Web, here, not the _American_ Web.)
Furthermore, your argument also says _nothing_ about how commercial Internet services where financing themselves back then which was the original topic of this thread. GeoCities, Yahoo, AltaVista, Google -- all those sites featured adds back in the 90s. If I remember correctly, Netscape Navigator also featured a "what's cool" button that -- I believe -- brought you to sites that payed for being advertised that way. So even back in the Internet's infancy it was largely commercialized using advertisements. This has never changed. And this was my original point.
I am not saying I am big fan of ads. But this is largely a problem of news agencies that work around a clock to bring you news (or rumors) and are currently unable to finance their services by subscriptions alone. This is the sector where ad spam is most visible. Despite all that the Internet has become a tremendously important part of our lives bringing real value: We get the news from there, we go shopping there, we watch live broadcasts there, watch movies, stay in contact with our friends, hold meetings, apply for jobs or housing and even work in there. The Internet was a toy in the 90s. These days it is central to our social lives. Saying the Internet is "broken" today and we have "skrewed it up" just because of its advertisement sector is a bit naive ...
I didn’t want to even get into this but yea as a 9 year old trying to set up IRC on an AOL connection by like 1996, this reeked of BS. I am only trusting my own hazy recollection here though, so thanks for validating it. Sometimes I feel like I am crazy describing how things used to be.
> Google's primary business model way back in 2000 was already showing ads related to your search terms. Even back in 2003 Google introduced the free GMAIL service that showed ads based on your email content. That's 21 years ago.
The adtech of today is nowhere near as invasive or pervasive as it was then - this is an extremely dishonest or ignorant framing of what is happening in today's internet vs the one of yesterday.
There are plenty examples of paid-for subscription services on the internet doing profitably without jamming adware/malware down your throat. Would you like examples?
I have paid maybe 10k in my lifetime to be connected to internet. That is why the telecom company has been able to invest in infra. I don't have good fiber thanks to Netflix.
https://joinsymbol.com — Manipulate text with translations, JSONs, and other revisions with AI. Provides an API for using in your apps, etc. CMS on top AI basically.
My current strategy is to go inch deep, mile wide. I don’t want all of my eggs in one basket any more. Basically doing a VC model using my portfolio of products.
Holler at me if you want to chat about these products or if you have any ideas you want to build.
Just "infinitely curious individual" is a bookworm. Hacker would also like to tinker with the rules underlying the systems to see how things might break. So bookworm + tinkering + wants to understand complex systems (hackers call it "to grok"). Systems - not only computer systems, also biological, physical and any kind of complex assemblages of rules.
Still not specific enough and doesn't evoke hackers. I consider myself a hacker but I don't consider myself "infinitely" creative, just creative in technical areas. So maybe technically creative individual? But those are more like Makers.
