I work for an AR/VR manufacturer. One of the jokes I used to hear all the time is that the market for VR is 10 years away, just like it’s always been. The success of various companies make it seem like we may finally be getting there, instead of it being a passing phase/gimmick.
I remember reading an interview with Jerry Garcia sometime in the late 80s waxing enthusiastically about VR. There were headsets even then that were impressing new users. As for when it’ll be 100% convincing, like Strange Days convincing, I’ll believe it when I see it.
I’ll throw another vote in for generative AI, though mostly on the art side (LLMs still feel like a search engine crossed with a parlor trick to me). We have the NightCafe thread in Cafe Society and the differences in results over the last 18 months are simply amazing – that much more so if you look at more sophisticated models/engines than NightCafe is using. Even jokes about mutant squid-hands are becoming rapidly obsolete.
I’m sure it’s had a long slow burn leading up to the point of these rapid advances but the last year and a half has really been astounding. Each time I’ve thought an advance is probably a long way off, it’s upon us in a few months. It’s gone from “amusing diversion, usually to laugh at” to “legitimate tool for image generation” in no time.
The dramatic increase in our online capabilities over the past 5 years amazes me. For a long time, the emphasis on technological advances was on hardware. Bigger hard drives, more memory, faster processors, etc was the order of the day. That has changed because virtually everyone has hardware that is more than capable of doing whatever it is they want to do. Now, with the help of COVID driven necessity, the emphasis is on online capability.
Flying cars are always 20 years away!!!
I was thinking generative AI. That, or the current capacity of Flash memory. Or petaflop (8 bit) GPUs.
To a certain extent I agree with Veep character Ben Cafferty: “I liked the internet better when it was just Alta Vista and that Star Wars kid!”
Which is to say I can do without a lot of the latest bells and whistles. I can’t think of much from the last five years that really matters to me. The last thing that really helped my day to day life was getting spoken driving directions from my phone. Before that it was the introduction of the iPad and the simplicity it brought to some of my work functions. And prior to that it was basic GPS.
Facebook used to be good tool for keeping track of friends and staying in touch… until they “improved” it by no longer using a linear timeline. Because it now functions best as a propaganda and misinformation tool a lot of my friends have left it behind.
WiFi is good, but it’s been around longer than five years…
Streaming movies? That was kind of good until the paradigm developed that you don’t actually own anything, and stuff you paid for gets taken down and becomes unavailable at the company’s whim.
A friend of mine showed me his fancy new VR headset. Kind of cool, but not something I would spend money on. Not until it begins approaching Star Trek holodeck capability.
Until this thread I hadn’t really considered it, but I think I could do without a great deal of new tech. A lot of it seems to work for its own benefit, not ours.
Yeah, for decades everyone was talking about how computers were going to allow people to go “paperless”. Yet in the 2000s I was printing out airline boarding passes, and event tickets, and MapQuest directions, and recipes from the internet, etc. It was getting a smartphone that allowed me to actually go paperless, as I now use my phone for all those things I used paper for a little over a decade ago.
The basic idea of neural nets has been around since the early days of computing, but had gone nowhere and more-or-less forgotten. Then–as I understand it–a team led by man named Yoshua Bengio at the University of Montreal decided to resurrect it and sufficient computing power finally existed to make it work.
IMO, the most astounding technical breakthrough in the past couple years has been hardware and software to allow people with severed spinal cords to recover function. It is very new, but I expect it to improve rapidly.
But again, not intending to nitpick, but computing power is a necessary but not sufficient condition for high-performance AI. The right architecture to effectively exploit it is also needed. Bengio is known for advancing techniques of deep learning, which require a multi-layered neural net architecture. The layers can be regarded roughly as the levels of abstraction in a typical communications protocol stack (except inverted) – that is, each layer learns to perform a well-defined operation on the data and passes the result to the next one which learns to establish more refined and complex relationships in the data set. Very broadly speaking, the idea of layers of abstraction has been used, in one sense or another, to architect solutions to otherwise intractably difficult problems.
As I understand it the important technological innovation behind GPT is the “transformer” architecture (the T) which was invented around 5 years back. The key paper was written in December 2017 and the first models like GPT1 and BERT were created in 2018. I suspect that even AI researchers from 6 years back would be stunned at what GPT 4 can do.
And ultimately it’s not just about the underlying technology but creating a service that is user-friendly and accessible and which becomes popular. I don’t think I have seen any technology product explode in popular consciousness and use as quickly as Chat GPT.
IMO, there’s only really been one large breakthrough in architectures in the last 20 years, the transformer. That paper was 2017. The rest, yeah, is mostly better data and better compute. Layer norm or residual connections or dropout are fairly minor is comparison. Richard Sutton’s Bitter Lesson is bitter for a reason; being clever hasn’t worked as well as scaling up. The biggest drivers in the advance of AI have been 1) internet scale datasets 2) GPU computing 3) transformers.
That said, generative AI blows my mind as well. When I first saw https://thispersondoesnotexist.com/ my jaw dropped.
I had been living without any decent internet for 30 years. Oh, a couple of satellite companies made it ‘possible’, and I dealt with it. No other options. None.
When COVID hit, Starlink came out. I was hesitant. A friend down the road put his dish up, and had no problem working from home.
It has been flawless. I work from home, stream all my internet, music, and television.
-
Generative AI and LLMs. No question this is number one. It might be number one for the past 30 years. It might even rank up there with the computer and the steam engine for global impact.
-
SpaceX Starship and reusability in general. A complete revolution in space access. A hundred years from now, space travel history will be broken up into the pre-reusable phase and the post-reusable phase.
-
Starlink. Global, inexpensive, high speed and low latency internet will be a revolution in the 3rd world and rural areas.
Repeating the wrong doesn’t make it right.
Just going to repeat a question I’ve asked before, that got no response:
Yes, Starlink is great for affluent ruralites like enipla. So you got that part right.
However, I do appreciate the irony of someone lauding supposed cutting-edge tech still using that dinosaur term, “3rd world”
Frederique Constant Monolithic watch. It’s a mechanical watch with a 40Hz movement and 80-hour power reserve, while the typical watch uses a 4hz movement and last 40 hours or so.
I assume that the price will vary depending on the local market conditions. So not $100 per month everywhere.
Quibbling about the exact figure is so beyond the point, but:
$100/month is what was being actually charged in the Indian pilot project. OK, strictly speaking, ₹7500 is $92/month. In a country where 75% of the rural population survives on < $80/month income, mind you.
Here in South Africa, it will roll out at over $110 - for the shittiest lower 50Mbps tier.
I’ve seen figures quoted for other African services that work out to more like $40-50. Rural Africa is, however, not as rich as rural India, the argument still stands, even at that price point.
Like I said in that other thread, most of y’all just don’t have clue one about what actual grinding poverty in the rural parts of the least-developed Asian or African countries is like. The cities not being that much better, mind you. $100/month covers a lot of people.
Anything more than $10/month does not qualify as “inexpensive” in that kind of situation.
I’m going to vote for generative AI, too. That feels like the next big revolution, and the one that may put a lot of my friends out of work, and help others be vastly more productive.
While i look forward to a foldable phone in my near future, i can’t imagine that will influence much more than the shape of the lump in my pocket.
And i played with an oculus quest during the pandemic, and remain underwhelmed. That’s a technology that will make a difference, but i don’t think it’s there, yet.
I’m going to throw in a vote for new techniques in vaccines. mRNA vaccines are truly new in the last 5 years, and recombinent viral vaccines didn’t really make it until recently. (The first was the ebola vaccine introduced in 2014-ish, in a large and successful test during an outbreak.)
The big thing with the foldable phones seems to be that the screen itself is able to fold. I imagine that you could almost do the same thing if there were two separate screens separated by a very small break, so I don’t see that foldable phones are that much of a big deal.