New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
What do you think about Situational Awareness (AGI By 2027 Prediction)
raindog308
Administrator, Veteran
in General
https://situational-awareness.ai/
AGI by 2027 is strikingly plausible. GPT-2 to GPT-4 took us from ~preschooler to ~smart high-schooler abilities in 4 years. Tracing trendlines in compute (~0.5 orders of magnitude or OOMs/year), algorithmic efficiencies (~0.5 OOMs/year), and “unhobbling” gains (from chatbot to agent), we should expect another preschooler-to-high-schooler-sized qualitative jump by 2027.
AI progress won’t stop at human-level. Hundreds of millions of AGIs could automate AI research, compressing a decade of algorithmic progress (5+ OOMs) into ≤1 year. We would rapidly go from human-level to vastly superhuman AI systems. The power—and the peril—of superintelligence would be dramatic.
Comments
apt-get install singularity(it's a real thing)
AGI would entail that it is self-aware of itself and would have independent thoughts and goals and opinions.
It also depends on what we consider "self aware".
Currently when you ask the AI what it thinks about something, it will combine the question based on the training data (absolute or average value) of what it has read from the internet, and displays it as answer. Human can do the same thing, but it doesen't mean that he will agree with the said results. Current chatgpt cannot form its own opinions, nor its "aware" as in "thinking". It simply combines topics it has been trained on with database-like structure and fixes the scentence to human readable text based on LLM guidelines. If you ask it if its aware its a machine, it basically checks from training data what machine is, checks the context such scentence is uttered, finds that it has been asked before, checks past answers, copy paste. Or it could have static value on itself where its defined as machine or non-living thing. Then it simply grabs that info and sends out message rather than "understand" you are asking about "its" opinion. It has no sense of self. Its just a huge ass database with parser. A search engine if you will.
Basically its stupid machine which just responds your inputs with combined information from training data and formatting which looks humanlike. Theres no point in the process where it "thinks", ohh I wonder what does this dude looks like or why am I even bothering to do this?
LLM alone wont make it happen, but can probly create such illusion, but something will always feel "missing". Probly needs another approach.
As for the Open-AI video about GPT-4o
That demo is combination of multiple independent things..
As for risk of rogue AI.
You simply isolate it. Even if it breaches the network hardware, you pull the plug from the hardware and it can no longer "influence" outside itself. Plus, this is why you don't give sensory organs to such "system". No matter how intelligent it is, as long as it cant influence us outside its own little box, we can just read and output everything it processes or "thinks" in one way.
Limit bandwidth so it wont "escape", suppose it could "superseed" itself out like in bittorrent, that would still take ages given how massive it is. Serve new data to it via local transfer methods, like physical drives. Have analog systems manage the main power feeds.
Even in case such an idiotic fantasy situation exists where theres AI so advanced it attempts to plot something, as long as you dont give it hands, legs, or worse, independent moving terminals with wireless connection, it cant do shit. As for nukes? They are not connected to internet. You have radio and paper code and physical person at the other end. Its closed circuit always. So the Skynet thing will never happen even if entire internet gets compromised by such super AI.
There was some religious nutjob dev at Google few years back who claimed it was self aware. This is same kind of fantasy.
AI will make my morning hot dogs before Biloh will fix the rules, that’s for sure.
This is the peak of AI which you should be afraid of.
Its called "The Orfice", and it will doom all of us.
https://www.reddit.com/r/orificeai/comments/1bzaxms/progress_update_for_orificeai_the_worlds_first_ai/
LLMs are the wrong approach and will plateau before anything resembling AGI
Language is a tool for defining consciousness, not consciousness itself. So-called artificial intelligence based on language will never have self-awareness because self-awareness is not based on language. Instead, language is based on consciousness. Humans cannot transcend language to express consciousness. For example, the human brain controls the body to perform various actions not through language but through intuition, which is highly energy-efficient. However, machines need to use predefined language to complete the same tasks, making it mechanical and requiring a lot of computation, thus consuming more energy. In fact, current artificial intelligence cannot even understand language; it generates text letter by letter based on probability calculations.