Jump to content

Welcome to FutureTimeline.forum
Register now to gain access to all of our features. Once registered and logged in, you will be able to create topics, post replies to existing threads, give reputation to your fellow members, get your own private messenger, post status updates, manage your profile and so much more. If you already have an account, login here - otherwise create an account for free today!

Community Status Updates


Photo

Yuli Ban

We're to hear of more details about the Kernel Flux on March 6th
Mar 06 2021 05:41 AM
  • Set and Meet Goals's Photo
    Set and Meet Goals
    Excited as fuck, please share your thoughts when new details are released
    Mar 06 2021 06:13 AM
Photo
Photo

Yuli Ban

By staggering chance, I'm in Houston. It's as bad as they say.
Feb 17 2021 06:59 PM
  • funkervogt's Photo
    funkervogt
    I definitely am.
    Feb 17 2021 08:07 PM
  • Yuli Ban's Photo
    Yuli Ban
    Waffle House is closed here. That's how you know.
    Feb 17 2021 08:33 PM
  • Cyber_Rebel's Photo
    Cyber_Rebel
    Been talking with family members down there so I have some idea, stay safe, and preferably warm if you can Yuli.

    And well, now you guys get a good New Yorker style winter. I remember the winter was what I hated most moving here being southern native myself.
    Feb 17 2021 08:37 PM
Photo

Yuli Ban

We should be hearing about the first exascale supercomputer sometime soon. Likely out of nowhere. Do you think it'll be Chinese, American, or Japanese?
Feb 06 2021 05:55 PM
  • TranscendingGod's Photo
    TranscendingGod
    I'd be willing to bank on Japan or the United States simply due to the fact that China appears to be a black box recently when it comes to news about their supercomputer plans. And so I'd give the United States and Japan about an even chance.
    Feb 06 2021 06:47 PM
  • Raklian's Photo
    Raklian
    We know there is high probability there will be a debut of the first exascale supercomputer sometime in 2nd half of 2021.
    Feb 06 2021 09:15 PM
  • waitingforthe2030s's Photo
    waitingforthe2030s
    Probably Japan, I heard something about them creating one a while back.
    Feb 07 2021 07:30 PM
Photo

Yuli Ban

The Big Sleep has taken over /r/MediaSynthesis. The days of consistent text-to-image are finally upon us https://www.reddit.com/r/MediaSynthesis/
Feb 02 2021 12:32 AM
  • PhoenixRu's Photo
    PhoenixRu
    Text-to-image? Let me guess, most of common users will (as always) use it to create porn (asking AI to visualise their dirty stories or something like that).
    Feb 02 2021 07:45 AM
  • Erowind's Photo
    Erowind
    woah
    https://www.reddit.com/r/MediaSynthesis/comments/la4z46/big_sleep_nostalgia/
    Feb 02 2021 11:33 AM
  • Yuli Ban's Photo
    Yuli Ban
    @PhoenixRu: You bet! That's what most AI Dungeon stories already are.
    Feb 02 2021 06:08 PM
Photo

Yuli Ban

Doubly AI generated! Took a scene from AI Dungeon and then Story2Hallucinated the scene into animation with The Big Sleep https://v.redd.it/2klpohbdvae61
Jan 30 2021 05:41 AM
  • Yuli Ban's Photo
    Yuli Ban
    This is what I've been talking about for years. It's only a matter of time before you need only write something and an AI synthesizes a whole movie out of it.
    Jan 30 2021 05:41 AM
  • Erowind's Photo
    Erowind
    https://www.youtube.com/watch?v=_SgWum9kHYk

    https://afinde-production.s3.amazonaws.com/uploads/2eee0567-1b6d-4553-bebc-358003433f17.gif
    Jan 30 2021 10:25 AM
  • Raklian's Photo
    Raklian
    Dreams - "I'm so weird and no one will ever be stranger than me."

    AI Dungeon/Story2Hallucinated - "Take my dream beer."
    Jan 30 2021 07:05 PM
Photo

Yuli Ban

Thanks to various bumps of old threads, I've gone around reading various 7+ year old threads and marvel at how optimistic and sci-fi oriented so many people were, especially considering the limitations of technology. At least on this forum, I've definitely noticed a shift away from sci-fi dreams towards the immediate future, among other trends.
Jan 19 2021 07:36 PM
  • Erowind's Photo
    Erowind
    God I hope FIVR is open source. I think here specifically a slew of new members, like 10-15 new active members would help so much. Most of us have already gone through the beats of dreaming of FIVR and AI already.

    A change in historical epoch would help too. Once we're seeing robots in our daily lives or AI assistants and AI generated media are widespread there will be more discussion too.
    Jan 20 2021 08:37 AM
  • tomasth's Photo
    tomasth
    I don't see any change from the "how optimistic and sci-fi oriented" , because of advances in media synthesis and scaled machine learning , there is more focus on its near term speculation.
    Jan 20 2021 01:09 PM
  • tomasth's Photo
    tomasth
    Does any one know if eric of KurzweiAI , is a member ? I'm curious of his opinion (became of a comment he once made).
    Jan 20 2021 01:12 PM
Photo

Yuli Ban

Fun fact! The word "robot" referring to an automaton was coined 100 years ago next week! https://en.wikipedia.org/wiki/R.U.R.
Jan 17 2021 07:16 PM
  • kjaggard's Photo
    kjaggard
    I read that book in 6 or 7th grade, i barely remember it, but I do recall feeling like they were using the term differently than we recognise it today. Weren't the robots in that more like bladerunners replicants? in that they were grown or something like that?

    Jan 17 2021 09:36 PM
Photo

Yuli Ban

Looks like we might hit 5,000 COVID-19 deaths per day in the USA either this week or next week
Jan 14 2021 02:12 AM
  • wjfox's Photo
    wjfox
    Horrific :(
    Jan 14 2021 08:04 AM
  • Zaphod's Photo
    Zaphod
    Things are very grim in the UK too. 1,564 deaths yesterday - a record and it looks to keep rising for a bit. This new variant has been brutal.
    Jan 14 2021 08:46 AM
  • Erowind's Photo
    Erowind
    Seconded :(
    Jan 14 2021 01:07 PM
Photo

Yuli Ban

It'll be interesting to see the reactions to some upcoming technology later this year. I wonder if certain skeptics will finally be moved...
Jan 11 2021 08:02 PM
  • waitingforthe2030s's Photo
    waitingforthe2030s
    What can you tell us?
    Jan 11 2021 08:31 PM
Photo

Yuli Ban

Reminder that we are one app away from musical style transfer (e.g. swapping instruments, genres, and vocalists): https://www.youtube.com/watch?v=YQAupr7JxNY
Jan 10 2021 12:54 AM
  • Yuli Ban's Photo
    Yuli Ban
    The technology is already here and has been for a few years now. It was already above-decent in 2018. The only thing keeping us from an actual application where you can take, say, a heavy metal song and turn it into a particularly angry-sounding 70s funk track is technical will.
    Jan 10 2021 12:56 AM
  • BrightWave's Photo
    BrightWave

    What about 'Voice Acting'?, Will real-time same day anime dubs be a thing in the near future? with half decent voice acting?
    Jan 10 2021 01:11 AM
  • waitingforthe2030s's Photo
    waitingforthe2030s
    It might already exist, but hasn't been released for obvious ethical or copyright reasons. Of course, someone's going to release something like it sooner or later.
    Jan 10 2021 03:55 AM
Photo

Yuli Ban

Seems very likely we'll have at least a zombie-AGI by 2025. This'll be almost doubtless by the end of the year.
Jan 09 2021 12:50 AM
  • Yuli Ban's Photo
    Yuli Ban
    Starspawn0 defined zombie AGI years ago: https://www.futuretimeline.net/forum/topic/18340-what-a-future-zombie-agi-might-look-like-and-how-it-might-be-built/
    I've been using the term "first-gen AGI" but I'm starting to come around to "zombie AGI" because I get what Starspawn0 meant by it
    Jan 09 2021 04:26 AM
  • Yuli Ban's Photo
    Yuli Ban
    To put it simply it's an AI that can pass a "true" Turing Test— a 30-minute or hour-long conversation with a human, any human, under any circumstances with zero handicaps (rather than the very short, highly curated, and highly handicapped test that Eugene Goostman passed in 2014). The skills required to do this are exquisitely challenging to make. There is no possible way to hand-code a computer to pass such a test, meaning it would HAVE to have some sort of semantic and symbolic...
    Jan 09 2021 04:31 AM
  • Yuli Ban's Photo
    Yuli Ban
    ... understanding of natural language as well as a multimodal set of experiences similar to what a human would have. By *some* accounts it could be called an artificial general intelligence. However, not all would agree and it would still have limitations. And by no accounts would it be human-level AI except *maybe* in terms of natural language generation. I can see GPT-4 passing a limited Turing Test, though not to the level of being called even a zombie or first-gen AGI.
    Jan 09 2021 04:33 AM
Photo

Yuli Ban

Novel video synthesis is closer than it appears.
Jan 08 2021 07:14 AM
  • Erowind's Photo
    Erowind
    it should be a matter of linking together 10 second gifs hopefully right? That can't be far away. (You probably wrote that on this forum somewhere Yuli but I can't remember where I heard the idea.)
    Jan 08 2021 09:46 AM
Photo

Yuli Ban

Can we go back to 2020?
Jan 06 2021 11:49 PM
  • Raklian's Photo
    Raklian
    LOL
    Jan 07 2021 12:07 AM
  • Cyber_Rebel's Photo
    Cyber_Rebel
    It's only been a week. 2020 was never going to be fixed in a week. Maybe a month or 2.... I hope.
    Jan 07 2021 05:35 PM
Photo

Yuli Ban

Here's a new way to think of synthetic media: not as computers creating things but rather "digital molecular assemblers." Pixels & bits are just like molecules— and AI is a molecular assembler.
Jan 02 2021 09:04 PM
  • Yuli Ban's Photo
    Yuli Ban
    They can't create long-form coherent stories more due to the fact that they learn entirely from text, which is an okay means of gleaning understanding the world but for the most part lacks the nuances of actual experiences that range from sights, sounds, tastes, temperatures, and so on. Further more, GPT-3 has fairly limited memory— its context window is only 2,048. It literally *can't* create a human-level short story with such a hobbled feature
    Jan 03 2021 08:19 PM
  • Yuli Ban's Photo
    Yuli Ban
    Simply adding multimodality and increasing the size of its context window (even by as little as 3x the raw number, though preferably much longer) would make the difference between GPT-3 and GPT-4 in terms of storytelling similar to the difference between a Markov chain and GPT-2.
    Jan 03 2021 08:27 PM
  • Whereas's Photo
    Whereas
    Well with how mind-blowing GPT-3 has turned out to be I'm certainly hoping for another leap in performance by GPT-4. What I'm getting at is that it's still not going understand what it's writing about ... though I suppose it might turn out that for many applications it doesn't really need to. For example would it really do a poorer job of covering a scientific paper than a journalist who *also* typically doesn't understand the subject matter?
    Jan 03 2021 10:08 PM