| Page 23 of 65 |
|
|
Posted: Thu, 23rd Jan 2025 00:18 Post subject: |
|
 |
Ai gonna cure cancer? 
|
|
| Back to top |
|
 |
|
|
Posted: Thu, 23rd Jan 2025 00:19 Post subject: |
|
 |
Last edited by Interinactive on Thu, 4th Dec 2025 07:13; edited 1 time in total
|
|
| Back to top |
|
 |
|
|
Posted: Thu, 23rd Jan 2025 01:27 Post subject: |
|
 |
Nhmm Norway uses Ai for oil drilling (Automation)
2. Key Technologies
Enhanced Oil Recovery (EOR): Techniques such as injecting CO₂ or water to maximize extraction efficiency.
Automation & Digitalization: Real-time monitoring, robotics, and AI are used to optimize operations and improve safety.
Carbon Capture and Storage (CCS): Norway is a global leader in capturing and storing CO₂ emissions from oil and gas production. For example, the Sleipner and Snøhvit fields have implemented CCS projects.
|
|
| Back to top |
|
 |
vurt
Posts: 14936
Location: Sweden
|
Posted: Fri, 24th Jan 2025 11:09 Post subject: |
|
 |
| Il_Padrino wrote: | | ASI is not going to come, not in a long shot. This is all money down the drain. |
Let's do AGI first, but i think we're already there, it's just that people who don't like AI keeps saying "but THAT's not AGI, it also has to be able to do X and Y!" So the goal post just increases all the time.
Super intelligence is not expected to come any time soon, but who knows. The advancements of AI is crazy, every week there's something new jaw dropping happening. There's already been many examples of self awareness. These models are so advanced that they're beyond human comprehension, we just grasp the foundational principles.
In 3 months, AI for coding has gone from ~millionth best coder to 175th best coder (competitive coding contest tests). Millionth best coder is top 2-3% in the world, so it's not like it's a bad coder at all, but 175th best is... yeah, it's pretty good.
The modern world basically runs on code, imagine everyone being able to use the top best coder in the world, unlimited amounts of this coder, and it works 24/7 and at lightning speed, at almost no cost. it's world changing, it's money very, very, well spent.
|
|
| Back to top |
|
 |
Il_Padrino
Posts: 8285
Location: Greece by the North Sea
|
Posted: Fri, 24th Jan 2025 12:52 Post subject: |
|
 |
I guess it depends on how one defines 'AGI', but in my opinion, 'pretty good' means it's not AGI.
AGI means the point technological singularity is reached, or very close by, with stand-alone, self-improving machines leading to unimaginable technological advances.
Not doing the same things a human could do, but (a lot) faster of course, which looks like the current goal of most if not all AI companies.
https://en.wikipedia.org/wiki/Technological_singularity
An other way I look at it, is simply comparing the AI models with the best 'AGI' we currently have, aka our own human brain.
The human brain weighs around 10kg and operates at an average of 0.12Wh (more or less a light bulb). A single ChatGPT query requires an average of 2.9Wh, and tons and tons of data centers and even nuclear power plants behind it to operate, and it still doesn't truly understand what it is writing. It's just not sustainable. The AI boom has only just begun, really, and we're already seeing signs of diminishing returns.
If we want to reach AGI with current AI models and algorithms, we are going to need a lot of currently not existing technologies (in energy, hardware, ...), that would each by themselves already be gamechangers for humanity.
I'd love to see it happening in my lifetime, but I'm afraid it's just a dream.
The numbers don't decide, the system is a lie. The river running dry, the wings of butterflies.
And you may pour us away like soup. Like we're pretty broken flowers.
We'll take back what is ours. One day at a time.
|
|
| Back to top |
|
 |
vurt
Posts: 14936
Location: Sweden
|
Posted: Fri, 24th Jan 2025 13:12 Post subject: |
|
 |
It's more than "pretty good" i would argue + the capability increase are just insane for every week that goes.
What AI was 3 months ago is seen as almost laughable.
How it will be in 2 years from now, when we are just NOW *really* getting into it, i don't think we even know. These previous years are the baby steps into AI, like 3DFX was the baby steps into 3D graphics.
I'm almost 100% certain i get to experience even Super Intelligence in my life time.
AI is getting cheaper and cheaper too (power). E.g chinese DeepSeek outclases ChatGPT we learned this week. Competition is like always great....
Not too much power is required to operate it, i can operate it (Deepseek) locally on a 3090, so no, you don't need a power plant.
Machine learning the models themselves though, that is what takes power and time.
|
|
| Back to top |
|
 |
Frant
King's Bounty
Posts: 24978
Location: Your Mom
|
Posted: Fri, 24th Jan 2025 14:50 Post subject: |
|
 |
| Il_Padrino wrote: | f we want to reach AGI with current AI models and algorithms, we are going to need a lot of currently not existing technologies (in energy, hardware, ...), that would each by themselves already be gamechangers for humanity.
I'd love to see it happening in my lifetime, but I'm afraid it's just a dream. |
AGI is entirely different from current AI. AI doesn't "understand" the data it's trained on, it's "just" very advanced pattern recognition and data processing.
AGI means highly cognitive abilities on a more general level and that it at some level can understand and process on a cognitive level. The general consensus among AGI researchers seem to be that we're decades away from true AGI. Until then we'll probably have better AI that to some extent can provide some similar albeit very limited/specialized semblance of true AGI.
As for AI, it's evolving at a break neck speed. It's like the industrial revolution: some key inventions/progress/knowledge unlocked the door to a technological explosion and a new age. AI is like a watershed moment for computing.
Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn!
|
|
| Back to top |
|
 |
|
|
Posted: Fri, 24th Jan 2025 16:52 Post subject: |
|
 |
Open AI newest (o3) model is rumoured to be able to improve on a level beyond the data it has been training on. Also it has beaten a test that is considered a benchmark for AGI. But beating a test isn't AGI. Source is Open AI though and they are desperate for publicity and funds, their buisiness is losing a lot of money:. Last week Chinese company Deepseek released an O1 ish model, open source. You can run that on a midrange GPU (wouldn't dare to do that, but still), open AI asks 200$ a month for that service.
Open AI is in serious shit imho. All the AI models are so close in performance and are all within 2-3 months of progress. Doubt it's a money making buissiness cause in the end someone will release their losing (but still awesome) ai model as open source.
Think it's pretty close though for AGI though, AI is developing at an insane speed:
may 2024 gpt 40, sept 2024 o1, jan o3.
AGI is also a matter of definition. Imho AGI should do every task at least as good as humans. So it should at least do office jobs as good as humans, that will take some time to achieve.
|
|
| Back to top |
|
 |
|
|
Posted: Fri, 24th Jan 2025 17:49 Post subject: |
|
 |
I (my personal stance on it) its not even remotely AGI until a part of its qualification to be so, is it can 'ponder' or rationalize on its on.
It can think on its own, and not just respond when it has input or manual stimulus and quietly idle in a 'void' otherwise.
Granted I don't define what AI intelligence and sentience is. but to 'match or surpass human cognitive capabilities' it needs to be able to have thoughts, even if rudimentary ones. Also it would need the ability to comprehend what it says, or is doing. Not just an 'advanced scrabble game words strung together that are weighted favorably to the ones it's given' or 'repeat a logical pattern of motions with reactionary compensation'. it needs to understand what it is doing.
And we are not even a hint of a whiff close to that in AI.
A robot washing dishes, vs a robot knowing what washing dishes is and the context of what that 'is' is needed. Otherwise, an automatic door could be stretched to say its AI as it 'reacts intelligently to stimulus and situational awareness of a dynamically changing environment to complete a task by assessing its surroundings'
|
|
| Back to top |
|
 |
|
|
|
| Back to top |
|
 |
zenux
Posts: 3338
Location: lɘɒɿƨI
|
Posted: Sat, 25th Jan 2025 11:03 Post subject: |
|
 |
@PickupArtist
| Quote: | Here is a summary of the transcript in 5–10 bullet points, organized by key timestamps and themes:
Introduction and Frustration with Influencer Culture (00:00–00:53):
The speaker introduces a video critiquing influencer culture, expressing disdain for its ethical shortcomings.
Mentions a prior critique of greenwashing by high-profile influencers like Marques Brownlee, who failed to ask critical questions during an Apple interview.
Ethical Failures in Sponsored Content (03:40–07:06):
Discusses how some YouTubers knew about scams by sponsors but failed to inform their audience, prioritizing self-image over transparency.
Criticizes Linus Tech Tips (LTT) for failing to disclose harmful practices of sponsors like Honey, a browser extension accused of unethical practices.
Manipulation and Victim Narratives (10:15–12:02):
Highlights LTT’s pattern of shifting focus from critical issues to personal victim narratives.
Points out LTT’s selective retractions and lack of accountability for promoting questionable sponsors.
Double Standards and Gaslighting (20:01–24:47):
The speaker accuses LTT of holding others to higher standards than themselves while engaging in manipulative practices, such as selective communication to control narratives.
Draws parallels to broader ethical lapses in influencer culture.
The iMac Incident and Exploitation of Content (31:37–39:24):
Critiques LTT for monetizing an iMac repair incident and using it as entertainment while downplaying the ethical responsibility toward collaborators.
Calls out the exploitation of smaller creators and manipulation of narratives for financial gain.
Misuse of Morality (47:16–48:53):
Condemns LTT’s use of “morality” as a weapon to deflect criticism while mocking consumer concerns through merchandise and rhetoric.
Highlights the missed opportunity to lead by example as an influencer.
Gaslighting and Narrative Control (53:37–55:22):
The speaker accuses LTT of deliberately misrepresenting communication with others, such as using outdated contact information to create false grievances.
Criticizes this pattern as calculated narrative manipulation.
Broader Commentary on Influencer Ethics (58:35–1:00:11):
Advocates for influencers to prioritize transparency, accountability, and the well-being of their audience.
Emphasizes the need for influencers to work together to champion ethical causes, such as consumer rights and repair advocacy.
This transcript critiques systemic ethical failures within influencer culture, using LTT as a case study for manipulation, double standards, and prioritization of personal image over audience welfare.
|
I got 2 AI chrome extensions for YT vids that both do subs transcription and chatgpt summary after .
I did the summary of the video using one of them or both, I can't be bothered ATM to make sure which one did the job.
Glasp Web Highlighter: PDF & Web Highlight 1.1.27
YouTube Summary with ChatGPT & Claude 2.09
I didn't watch the video nor read the transcript or the summary.
|
|
| Back to top |
|
 |
|
|
Posted: Sat, 25th Jan 2025 14:06 Post subject: |
|
 |
but ai cant summarize whats happening in the video if the substitles dont mention it right ? like showing a screenshot or opening another video inside the video
|
|
| Back to top |
|
 |
vurt
Posts: 14936
Location: Sweden
|
|
| Back to top |
|
 |
zenux
Posts: 3338
Location: lɘɒɿƨI
|
Posted: Sat, 25th Jan 2025 15:24 Post subject: |
|
 |
| PickupArtist wrote: | | but ai cant summarize whats happening in the video if the substitles dont mention it right ? like showing a screenshot or opening another video inside the video |
Correct. I think, AI ATM can only summarize transcripts on YT. I've just noticed, all the vids I try to watch, that have people talking in them, come with YT provided auto-generated transcript.
Vids with no spoken speech lack transcript. I checked a dozen languages, all have transcripts in their respective languages.
Edit: checked a little more, some vids don't have transcripts even thou spoken words are present, such vids can't be summarized by available to me free tools. They offer me this in this case:
"No Transcription Available... 😢
Try video transcriber on Hugging Face
or Google Colab to transcribe the video!"
|
|
| Back to top |
|
 |
|
|
Posted: Sat, 25th Jan 2025 17:53 Post subject: |
|
 |
and will it just use that single source of info, or add stuff on its own
for example if i asked it to summarize why the 4090 is better then 4080 gpu from a youtube review video. or will it start using other data outside the video ?
|
|
| Back to top |
|
 |
|
|
Posted: Sun, 26th Jan 2025 08:36 Post subject: |
|
 |
This is a thing I think is realy scary considering lastest AI developments
-Close to state of the art AI models are released as open source now and will be in the future (cause the loser AI models will be released open source)
-AI models can self improve in the near future (this is close to current AI abilities)
-3000$ ish will get you a decent AI rig.
So AI will be very decentral and everybody can run it's own. It will spread like a virus and the weirdest shit can develop.
People always believed that one "magical" AI model would take over, but what will get is a swarm of self improving AI's running on locally owned machines.
|
|
| Back to top |
|
 |
|
|
Posted: Sun, 26th Jan 2025 08:45 Post subject: |
|
 |
@friketje
Are you forgetting local PC's have a power button?
It needs power and needs internet. Both of which has zero control over if it has or not.
AI is going to take over from locally run (that you need to manually run and leave running) only if people forget how to simply turn something off with a switch.
Now if AI can override my power button, and stop me from unplugging it..I guess. Otherwise kinda hard to imagine local home ran AI swarming the world when all you have to do tell people to push one button to stop it...
|
|
| Back to top |
|
 |
Il_Padrino
Posts: 8285
Location: Greece by the North Sea
|
Posted: Sun, 26th Jan 2025 08:57 Post subject: |
|
 |
Unless people fall for its emotions and pleas.
"What are you doing, Dave?"
The numbers don't decide, the system is a lie. The river running dry, the wings of butterflies.
And you may pour us away like soup. Like we're pretty broken flowers.
We'll take back what is ours. One day at a time.
|
|
| Back to top |
|
 |
vurt
Posts: 14936
Location: Sweden
|
Posted: Sun, 26th Jan 2025 10:12 Post subject: |
|
 |
The biggest threat with AI's is the increase in scams, it can be used on all fronts, used for hacking etc. We will go from quite a few can do it to anyone can do it. On the other hand, maybe we can also use AI to strengthen security?
I was thinking the other day how we can perhaps become more unplugged instead of more plugged in.. like grabbing big amounts of data quickly (e.g all the latest youtube videos in your subs, or your favorite web pages), and then unplug, mainly use internet as a form of updating in short bursts, every now and then... obviously this doesn't work for online gaming or anything that needs a continuous usage.
|
|
| Back to top |
|
 |
|
|
|
| Back to top |
|
 |
|
|
Posted: Sun, 26th Jan 2025 14:28 Post subject: |
|
 |
Brad Pitt scammer who are these people. I'm now scarred of Joe Pesci scammer
Brad Pitt is sick? Where can I send money 
|
|
| Back to top |
|
 |
|
|
|
| Back to top |
|
 |
|
|
Posted: Sun, 26th Jan 2025 18:36 Post subject: |
|
 |
| SumZero wrote: | @friketje
Are you forgetting local PC's have a power button?
It needs power and needs internet. Both of which has zero control over if it has or not.
AI is going to take over from locally run (that you need to manually run and leave running) only if people forget how to simply turn something off with a switch.
Now if AI can override my power button, and stop me from unplugging it..I guess. Otherwise kinda hard to imagine local home ran AI swarming the world when all you have to do tell people to push one button to stop it... |
That's not the point. It's the idea that everyone with basic computer skills can cook it's own AI model that is also able to self improve or alter itself in different ways. No one will have control about how it develops and an insane amount of different AI models will develop.
And considering power buttons: the internet is always on. And people with bad intentions will use powerfull AI as soon as it's available. Considering rogue AI: the more models, the bigger the chance something weird will happen.
|
|
| Back to top |
|
 |
|
|
Posted: Sun, 26th Jan 2025 19:00 Post subject: |
|
 |
I think you overestimated the 'basic computer skills' aspect. Most people don't have 'basic computer skills' now.
Working with my circle of people, and having a best friend that runs a PC shop. You would be amazed the percentage of people that's skills stop at: It's not on my desktop, I guess it's lost forever.
All your scenarios require best-case 'hypothetically possible' situations.
I see it as the same as 20 years ago "The internet will open to every person all the knowledge of mankind at their fingertips." Yea, that was hypothetically possible. But what did it become? People watching cat videos, nut-shot videos, and posting toilet thoughts on twatter.
And just like the internet, and computers, and more complex intranet situations so will be AI. people with bad intentions will use it to try to do bad things, and people will good intentions will use it to mitigate that.
And either way, we need to travel the 1000 miles to an AI having even a single basic 'thought'. One day maybe? Sure. But a single-cell creature has exponentially more self-decision power than an AI (As even 1 of them is infinitely more than the 0 AI has).
The AI and its form we have now, isn't even structured to be able to evolve into that. It's not a path it can take. We'd need a whole new, ground up, type of AI to even be able to move down that path. it's whole foundational structure is wrong to even attempt to make it do that. It's like trying to 'teach' a house to be an airplane. (Its basic formation is not able to go that direction to ever fly).
In your scenario, you are already granting AI free will and the ability to think on its own, and have reasoning, motivations, intentions, and self given purpose. Those are tall orders we arent even starting to fulfill, nor ever can with the way AI is now. Again, we'd need a whole new dynamic of a totally different AI from scratch to 'add that feature'
|
|
| Back to top |
|
 |
|
|
Posted: Mon, 27th Jan 2025 16:00 Post subject: |
|
 |
|
|
|
| Back to top |
|
 |
|
|
|
| Back to top |
|
 |
Il_Padrino
Posts: 8285
Location: Greece by the North Sea
|
|
| Back to top |
|
 |
|
|
Posted: Mon, 27th Jan 2025 20:26 Post subject: |
|
 |
It's a bit early to jump to conclusions. Also doubt the Chinese did this with 6 million dollar, wouldn't be surprised they've "borrowed" some foreign AI model.
Markets don't care. It's not about the actual news, but just an excuse to sell off stock for a quick buck.
Open AI has a problem though, doubt there is a lot of money in AI models cause every company doing a serious effort is able to keep up in the AI race, the models are worth shit.
Big tech and the semi buissiness will be fine though. Buissiness want cloud AI and more GPU's are needed anyway.
|
|
| Back to top |
|
 |
zenux
Posts: 3338
Location: lɘɒɿƨI
|
|
| Back to top |
|
 |
|
|
Posted: Mon, 27th Jan 2025 23:09 Post subject: |
|
 |
We've had wind turbine projects here and then some American data center comes over.
|
|
| Back to top |
|
 |
| Page 23 of 65 |
All times are GMT + 1 Hour |