Thiraviyam
Well-known member
- Joined
- May 22, 2023
- Messages
- 286
- Points
- 63
AGI is possible with quantum computing! When LLM s and NLP use transformers to predict and create its own content I guess AGI is not a impossible realm given the speed at which technology is growing!False. AI not even closer to be become sentient. It might not go that way, even if it goes it would still be be a trained Model by Humans. Even to use pre trained models these days requires huge computing powers.
Its all Hypothesis for now. LLM and NLP doesnt predict its own content, they use transformer architecture and datasets are regularly trained in a way to respond to the query. Again they are Trained, they don't think on their own.AGI is possible with quantum computing! When LLM s and NLP use transformers to predict and create its own content I guess AGI is not a impossible realm given the speed at which technology is growing!
I dunt say it will think! I'll rather say it learns and creates its own content, though the content is a human trained one! While Deepminds alphago AI can beat a world no 1 player at Go by learning its own moves I dunt think it uses any tokens and trained datasets,after certain moves 2 be precise! Alexa would ve been a sorcery item had it existed two centuries ago!Its all Hypothesis for now. LLM and NLP doesnt predict its own content, they use transformer architecture and datasets are regularly trained in a way to respond to the query. Again they are Trained, they don't think on their own.
While Deepminds alphago AI can beat a world no 1 player at Go by learning its own moves I dunt think it uses any tokens and trained datasets,after certain moves 2 be precise!
Content might not be human trained, but the accuracy of responses i.e accuracy of content response are human trained.I dunt say it will think! I'll rather say it learns and creates its own content, though the content is a human trained one
Yeah but what matters more is the accuracy keeps increasing and that's d catch!Content might not be human trained, but the accuracy of responses i.e accuracy of content response are human trained.
You should look at Google and Microsoft or even some other Big techs that frequently have competitions for designing AI models that can be more accurate and giving away millions of dollars as prize money. The biggest catch is most of these trained models and Frameworks are Open sourced lol.Yeah but what matters more is the accuracy keeps increasing and that's d catch!
Yeah be it Alphagos CNN or Google's RNN or LSTM every Model uses Algorithm patterns and statistics and probability to get fine tuned!but my question is what will happen when all of these models gets combined! As of now we dunt ve that computation power but with the advent of quantum computing I guess we might reach that potential!View attachment 14228
If you could understand the above diagram you will understand what AI like AlphaGo does. It still used Machine learning and Deep learning(neural network). Tokens aren't needed for these kind of AI since they don't do NLP, however they do get trained regularly to improve to where they are. AlphaGo also uses CNN which is primarily used most AI and especially machine learning these days.
They need datasets to get trained! I dunt guess it won't be possible unless untill it's open source! But lately OpenAi is at huge loss due to GPT I guess! Microsoft has been out of the league with Sam Altman! Bing has more access to GPT and Dall-E architecture than openAi I suppose!You should look at Google and Microsoft or even some other Big techs that frequently have competitions for designing AI models that can be more accurate and giving away millions of dollars as prize money. The biggest catch is most of these trained models and Frameworks are Open sourced lol.
If some country are willing to invest money and time into it, pretty sure other countries will find it unethical. its still an evolving field. Breaking barriers is gonna take atleast a decade.my question is what will happen when all of these models gets combined!
Given that exponential growth John McCarthy has envisioned it will happen around 2040 sumthing!If some country are willing to invest money and time into it, pretty sure other countries will find it unethical. its still an evolving field. Breaking barriers is gonna take atleast a decade.
OpenAI is profiting as many use their API for whatever GPT someone develops.OpenAi is at huge loss due to GPT
Microsoft are the biggest investors for the Open AI project hence they are benefiting. And that is why they implemented GPT into Bing.Microsoft has been out of the league with Sam Altman! Bing has more access to GPT and Dall-E architecture than openAi I suppose!
Yeah they invested more but what I heard is chatgpt is at loss! And so does Microsoft its biggest shareholder in terms of investment!OpenAI is profiting as many use their API for whatever GPT someone develops.
Microsoft are the biggest investors for the Open AI project hence they are benefiting. And that is why they implemented GPT into Bing.
They are at loss from last years revenue, pretty sure reaching a breakeven will take time. Microsoft invested close to $10billion. even for GPT the operation cost is generating loss and thinking quantum computing is pfft...Yeah they invested more but what I heard is chatgpt is at loss! And so does Microsoft its biggest shareholder in terms of investment!
ChatGPT creator OpenAI is reportedly earning $80M a month
And its sales could be edging high enough to plug its $540 million loss from last year.fortune.com