1 minute ago, Capris Groove said:

ChatGPT is what Google was to people in 2000. It's here to stay, but just as a tool.

We’ll see. The costs with serving these AI models is far greater than the cost of serving Google Search. For context (I’m working off memory so I might get the numbers slightly off), OpenAI brought in $14 billion in revenue last year. Its operational costs were $140 billion. These AI companies (and even the conventional tech giants like Microsoft) are bleeding money trying to find a way to hold onto users and turn these models into a viable revenue stream. But they can’t. Search Engines never had this type of downward spiral. Their business model was very straightforward and the cost to maintain them was nowhere near what AI models are in the infancy stage.

AI is “here to stay” sure. But AI has arguably always existed since the introduction of digital computers. The idea that LLM powered chatbots have no risk of going extinct… I think that’s not a foregone conclusion. These models cost too much and 95% of companies have found no ROI from using them.

8 minutes ago, Bloo said:

We'll see. The costs with serving these AI models is far greater than the cost of serving Google Search. For context (I'm working off memory so I might get the numbers slightly off), OpenAI brought in $14 billion in revenue last year. Its operational costs were $140 billion. These AI companies (and even the conventional tech giants like Microsoft) are bleeding money trying to find a way to hold onto users and turn these models into a viable revenue stream. But they can't. Search Engines never had this type of downward spiral. Their business model was very straightforward and the cost to maintain them was nowhere near what AI models are in the infancy stage.

AI is "here to stay" sure. But AI has arguably always existed since the introduction of digital computers. The idea that LLM powered chatbots have no risk of going extinct… I think that's not a foregone conclusion. These models cost too much and 95% of companies have found no ROI from using them.

It's difficult to imagine the technology simply ceasing to exist when what you're describing is a business model problem. Blockbuster went under but the technology (at-home entertainment) remained, same with how Spotify replaced CDs (and even Napster and Limewire, which were free).

I think LLM chatbots may also become a subscription model, and I think a lot of people would pay for them if the price was fair.

Edited March 28Mar 28 by Capris Groove

1 minute ago, Capris Groove said:

It's difficult to imagine the technology simply ceasing to exist when what you're describing is a business model problem. Blockbuster went under but the technology (at-home entertainment) remained, same with how Spotify replaced CDs (and even Napster and Limewire, which was free).

I think LLM chatbots may also become a subscription model, and I think a lot of people would pay for them if the price was fair.

I mean, LLMs aren’t “new”. Language models have existed for decades. So, yeah, it’s not going away. But the line between what’s always been around and what laypeople consider “AI” is blurry.

As for LLM subscriptions, it’s hard to say because they’re not transparent about the cost/revenue. But, I personally doubt they’ll get enough revenue to stay afloat. Google is losing a ton of money, as is Anthropic despite their success in being the “LLMs for working professionals”. I saw one report (though I cannot find it now so take it with a grain of salt), that suggested OpenAI’s subscription that was then around ~$200 was still a net loss.

1 minute ago, Bloo said:

I mean, LLMs aren't "new". Language models have existed for decades. So, yeah, it's not going away. But the line between what's always been around and what laypeople consider "AI" is blurry.

As for LLM subscriptions, it's hard to say because they're not transparent about the cost/revenue. But, I personally doubt they'll get enough revenue to stay afloat. Google is losing a ton of money, as is Anthropic despite their success in being the "LLMs for working professionals". I saw one report (though I cannot find it now so take it with a grain of salt), that suggested OpenAI's subscription that was then around ~$200 was still a net loss.

I wonder if they'll move to a subscription-only model at some point.

I want to erase its existence from my memory tbh. Not bc it can't be useful but bc it has now made me question EVERY SINGLE piece of visual or audio media I come across to. Is this ad AI? What about this poster? Or the lyrics of this song I like? It's driving me crazy. The thought alone takes a lot of joy away.

AI summaries can be very helpful but that's about it. Generative AI is ugly as **** and an energy/job harvester. Tired of AI slop

Generative ai has amazing purposes in the following:

• 3D modeling of isomers as well as pharmaceuticals with our known laws of chemical physics

• Predictive modeling for geospatial tracks and urban planning

• Building and simulating models of predictive patterns of things such as weather based on historical input

It is not useful or meant to be used for things like making ads. It's super counterproductive and a waste of resources. It's meant for complex problems and to reduce the amount of time that permutations of solutions take. Also, it's completely different from procedural AI, which is already used for things like driving and assisted surgeries. Those are actual practical solutions.

Here's the biggest issue with AI: it solves problems. How can you continuously charge for services when you solve the problem and reduce issues instantly? That's what big tech is struggling with. How do you make AI, which is the future, able to become a subscription?

I think it's useful but the industry will collapse under its own hubris.

I love it. I finally got both my idols to have a selfie together

I like AI fruit love Island

3 hours ago, Bloo said:

We'll see. The costs with serving these AI models is far greater than the cost of serving Google Search. For context (I'm working off memory so I might get the numbers slightly off), OpenAI brought in $14 billion in revenue last year. Its operational costs were $140 billion. These AI companies (and even the conventional tech giants like Microsoft) are bleeding money trying to find a way to hold onto users and turn these models into a viable revenue stream. But they can't. Search Engines never had this type of downward spiral. Their business model was very straightforward and the cost to maintain them was nowhere near what AI models are in the infancy stage.

AI is "here to stay" sure. But AI has arguably always existed since the introduction of digital computers. The idea that LLM powered chatbots have no risk of going extinct… I think that's not a foregone conclusion. These models cost too much and 95% of companies have found no ROI from using them.

these companies make AI to create a surveillance state for governments, so governments can monitor, influence and control people. as long as governments keep funding these companies they are fine

Edited March 28Mar 28 by reymiu

3 hours ago, Bloo said:

I mean, LLMs aren't "new". Language models have existed for decades. So, yeah, it's not going away. But the line between what's always been around and what laypeople consider "AI" is blurry.

As for LLM subscriptions, it's hard to say because they're not transparent about the cost/revenue. But, I personally doubt they'll get enough revenue to stay afloat. Google is losing a ton of money, as is Anthropic despite their success in being the "LLMs for working professionals". I saw one report (though I cannot find it now so take it with a grain of salt), that suggested OpenAI's subscription that was then around ~$200 was still a net loss.

You research AI for a living so I'll give the floor to you, but I'll just say that it's extremely rare for a technology which is already used by a huge portion of society to become limited in any notable way, unless it's replaced by something better. The last time that even happened was Damascus steel in the 1700s. So yeah, LLM access might be tiered, it might come bundled with other services, or individual companies (even big ones like Google) might go under, but on the flip side you might get open-source, other companies being created, etc. We'll see.

Edited March 28Mar 28 by Capris Groove

First let me state that I'm a goo-goo-gaga brained dumbass, so I have literally no idea what I'm talking about

But I think it's gonna kill us all. Maybe not within our lifetimes , but I think eventually it will be smart and autonomous enough to just nuke the entire planet just like the movies foretold. But again, what do I know ¯\_(ツ)_/¯

1 hour ago, KnightOfAllRealms said:

I like AI fruit love Island

You couldn't waterboard this out of me.

I'm fine with AI but also aware of the fact it's draining our water but also have accepted it's here to stay regardless.

40 minutes ago, TaylorsSquad said:

You couldn't waterboard this out of me.

I'm fine with AI but also aware of the fact it's draining our water but also have accepted it's here to stay regardless.

Candy love Island is even better sis I'm a stan of kween Chocolatina

From a data science standpoint, current AI models are very impressive. In many papers I've read and in my own AI publications, they are very promising in their abilities to answer questions in genome science. I want a future where precision medicine is a standard in healthcare, patients can get personalized treatments and prevention techniques, and we can develop treatments en masse due to drug target gene prioritization enabled by AI (or whatever its machine learning descendants are) and massive collections of data.

However, they are nowhere near as perfect or omniscient as Sam Altman would like to suggest. Some "impressive" applications of AI still only have like 60-70% accuracy. In some publications I've read, plain ol' statistical methods are still outperforming AI models.

With that being said, I would have preferred if AI was just kept as tools for data scientists. And that's not me trying to gatekeep, but the way it has infiltrated so many facets of our lives is nauseating. I get so many banal AI-written emails from university staff. Bonus points if they come with shitty images too. I may as well correspond with a brick wall.

Children and young adult brains were already fried from the constant screen time, social media algorithms, and Gen X helicopter parents micromanaging their lives. Now, the kids I meet are even more helpless because they don't think critically for more than 1 second without asking AI. People are either being deceived by AI-generated content or on edge about it. In this sense, the future is bleak, because in many ways, reality seems so blurry now.

With just my luck, the AI bubble will pop just when my dissertation is done and it's time for me to get a job. Hopefully at least one tech company will still be hiring. Maybe they can let me mop the floors. But I think it'll be for the best. AI isn't going away, but hopefully the attempts by these AI companies to propagate AI into every facet of our lives will abate as they bleed money, and hopefully humanity can correct course and integrate AI more carefully into our society.

11 hours ago, Bloo said:

It is a literal fact that a single inference call of an LLM (roughly 3–5 Wh with GPT-4) uses more energy than a search engine query (roughly 0.3–1.0 Wh with Google Search). AI generative models that make photos or videos consume even more energy. This is a fact. This is not up for debate. Further, you cannot say "well most of the energy waste is due to training not inference" to side step the energy costs. You cannot have a useful AI model to do inference without training. So, the cost of training has to be factored in.

3-5 Wh is not significant in the grand scheme of how much energy we use on things. You're gonna use more than that just by watching a movie, or (by a drastic amount) playing a video game, let alone using your devices at all. This sorta thing feels like fearmongering to me

Edited March 28Mar 28 by Kurt Hummel

4 hours ago, Kurt Hummel said:

3-5 Wh is not significant in the grand scheme of how much energy we use on things. You're gonna use more than that just by watching a movie, or (by a drastic amount) playing a video game, let alone using your devices at all. This sorta thing feels like fearmongering to me

It seems you have no sense of scale. Do you think people just use 1 inference call when they’re using ChatGPT..? Are you thinking of the hundreds of millions of people using these platforms at once? It seems not.

I’m telling you as someone that works in AI, the energy footprint is substantial. Saying it’s fearmongering for people to voice concern about the energy cost to produce videos with disinformation to confuse the masses is beyond bizarre.

Pictures-wise: I've been able to create photosessions for myself I always wanted. To correct old pictures of myself. To create pictures-concepts I always wanted. To make artworks of my favorite singers for my albums the way I want them. To test how fashion and hairstyles look on me. To test all types of art styles, and I have learned so much about art and graphic design.

Language Wise: I'm far more participative here-online because it has helped me fix my bad grammar-bad english

Music-wise: AI covers allowed me to understand how the English pop girls sound to native speakers. AI covers are usually boring, but a small number of them are really good. AI Music Software also allowed me to modify some songs the way I always wanted—a dream.

Education: I have so many questions about my homework and uni stuff, and AI has helped me to learn so many concepts as an assistant. It allowed me to clarify things. I wish I had this when I was in high school. I struggled so much, and my teachers were mean, and I needed something like this to help me learn.

Life: With AI, I can ask any questions about things in the world. I can upload any image and identify what is there, to identify species. We have a modern Pokémon Dexter. With AI, if I have questions about nature or any random thing, I have them in my hands. I can do deep research about so many topics I want.

I'm sorry, but to me, AI is everything. It has changed my life for the best, and I don't know how I could go back to living a life without it. I love technology and science above everything. I don't see it as something bad. I live for innovation and the future. I think there are ways for technology to work with nature and organic life.

Now when it comes to activism, I think people can be overly hateful and annoying with the AI slop thing. I also don't believe in the environmentalist ones because they live in some of the most consumerist wasteful countries in the world—buying fast fashion, wasting food, traveling all the time in airplanes, buying expensive merch from their faves and all types of items—a culture of waste and overconsumption. So the ethical thing looks sometimes more performative to me.

At the same time, I'm so concerned about the loss of jobs, and I feel really bad about the people suffering from it. It's like having the best and having the worst. But I'm not that concerned about AI itself. I'm more concerned about how we are led by evil techno pigs with no human empathy. People like Zuckerberg, Elon Musk, Altman, Thiel, Bezos, Gates—I can't believe all our leading tech heads are so sick. They should be ethical, highly empathetic, and sweet. They are from the best universities and were given big advantages in life. Goverments are also evil and not protective of people

Even with my love of AI, if I had a chance to rewind and change things in the 2020s, I would. I would go back to the early 2010s again, lol. No trump. no ai. no political bodegas. no wars. no global warming. no covid. .no bad bunny.domination. NO TO EVERYTHING
Yes to Gaga,Katy,Kesha,Rihanna recession pop

Edited March 28Mar 28 by AvadaKedavra

12 hours ago, qurl said:

First let me state that I'm a goo-goo-gaga brained dumbass, so I have literally no idea what I'm talking about

But I think it's gonna kill us all. Maybe not within our lifetimes , but I think eventually it will be smart and autonomous enough to just nuke the entire planet just like the movies foretold. But again, what do I know ¯\_(ツ)_/¯

I feel the same. Even if AI never thinks for itself like a human and only follows instructions, it would only take someone malicious enough to configure the code and algorithms of a Super Godly AI for it to become like a kind of robotic virus—self-replicating and willing to destroy all human life simply because it's in its programming.

Things like that make me question having children sometimes.