Just interesting. I'm not an expert but think either won't or will but not soon. With LLMs hype there seems to be a trend to tune models for very specific use cases and tuning is still up to human judges

Maybe someone can suggest beginner-friendly materials to read

1 year ago

Comment has been collapsed.

Its not very AI if it would process it as a human would.

1 year ago
Permalink

Comment has been collapsed.

Humans are inefficient

1 year ago
Permalink

Comment has been collapsed.

A human on a 2000 cal diet consumes as much energy as a light bulb. When they build a machine that can independently analyze enough data stored in itself to answer all the questions an elite polymath could (while consuming as much power as a light buld), then you can say humans are inefficient.

1 year ago
Permalink

Comment has been collapsed.

Ok, very closely simulate with no hallucinations

1 year ago
Permalink

Comment has been collapsed.

I for one welcome our new AI overlords.

1 year ago
Permalink

Comment has been collapsed.

Maybe it can set this crazy world straight 🤞

1 year ago
Permalink

Comment has been collapsed.

Well, an AI president would be better than a corrupt one.

1 year ago
Permalink

Comment has been collapsed.

What a nice one liner, Im stealing it for IRL purposes. Thanks😇

1 year ago
Permalink

Comment has been collapsed.

its a good idea to have it on internet record that i too love and trust ai

1 year ago
Permalink

Comment has been collapsed.

Will AI process data like a human brain?

ChatGPT
The processing of data by AI systems is fundamentally different from how the human brain processes information. While AI systems, including machine learning models, can analyze and make predictions based on large amounts of data, they do so using algorithms and mathematical computations.

The human brain, on the other hand, processes information through a complex network of neurons that communicate with each other through electrochemical signals. It has the ability to perform tasks such as pattern recognition, abstract reasoning, and creative thinking, which are still challenging for AI systems to replicate fully.

While AI has made significant advancements in various domains, it is important to note that current AI technologies are limited in their ability to replicate the full range of human cognitive processes. However, researchers continue to explore and develop new approaches, such as neuromorphic computing, that aim to emulate certain aspects of the human brain's processing capabilities.

1 year ago
Permalink

Comment has been collapsed.

No, never.

This is an article I found a number of years ago that I really like. It's about how we use metaphors to describe our experiences, and then get confused between the metaphor and reality. During the Greek times, when they were developing complex irrigation systems, they assumed the human body worked as a balance of liquids, the four humours. Later, during the enlightenment, they assumed human brains worked like the finely tuned bits of clockwork that were the peak of technology.

https://aeon.co/essays/your-brain-does-not-process-information-and-it-is-not-a-computer

1 year ago
Permalink

Comment has been collapsed.

Thanks for the article, sounds reasonable
IP abstraction used for human brain is actually popular and widely used
People might need tons of abstractions to explain and reason something

1 year ago
Permalink

Comment has been collapsed.

Not until we've found out how a human brain does it.

1 year ago
Permalink

Comment has been collapsed.

nope
to errar is human

1 year ago
Permalink

Comment has been collapsed.

We still have no idea how a human brain processes anything so it's unlikely we can replicate it in a machine.
Also "AI" is the new "VR". Sounds cool and all and the hype won't die because it gives people with column inches to fill something to say but there's no more Artificial Intelligence at this point than there is Virtual Reality.
One is a chat bot, the other is a couple of tvs strapped to people's faces.

1 year ago
Permalink

Comment has been collapsed.

No. While LLMs can generate text and simulate human conversations (in a convincing manner), it is not thinking or understanding anything like a human. Chatbots are not sentient nor intelligent, it is just us projecting anthropomorphism on a text generator...

Think of an LLM as a very sophisticated form of auto-complete. The reason LLMs are so impressive is because their training data is basically the entire internet.

1 year ago*
Permalink

Comment has been collapsed.

tbh i would suggest you to do a beginner courses on AI offered by deeplearning ai on coursera.. i have done it and doing an advanced courses on deeplearning atm.. its quite fascinating tbhj

1 year ago
Permalink

Comment has been collapsed.

Maybe in 50-80 years? I'm sure before then no one considered an artificial intelligence was even possible, and even though I'm not sure the ones today can be considered AI, technology is always growing.

1 year ago
Permalink

Comment has been collapsed.

I wouldn't ask about "like human", but about "as efficiently and universally as humans". The critical mass has probably already been reached and AI development increases exponentially, so I would say it's more like 5-10 years (closer to 10, than 5, IMO) before AI becomes intellectually clearly superior to us... the question still stands, will it then become self-aware and fully self-guided, i..e. a separate being living alongside us... or without us (Eliezer Yudkowsky: Dangers of AI and the End of Human Civilization | Lex Fridman Podcast #368 - https://www.youtube.com/watch?v=AaTRHFaaPG8).
My bet is, that AI thinking "like humans" will probably shortly follow, but it will not be of our design, but it will be the AI itself designing an inferior to itself "human simulator"... unless it decides, that we are so outdated, that it's not worth to run such a simulation ;(.

1 year ago*
Permalink

Comment has been collapsed.

I recommend you to take "AI for Everyone" by Andrew Ng. AI/NN uses complex mathematics to simulate human brain neuron activations.More n more work is being done(and more complexities are being added) but NNs perceive data very differently from humans......and will never be able to perceive or process data like human beings.

1 year ago
Permalink

Comment has been collapsed.

Thanks
I'm currently into building pipelines in a crowdsourcing data labeling sphere so this one looks useful

1 year ago
Permalink

Comment has been collapsed.

Happy to be helpful.

1 year ago
Permalink

Comment has been collapsed.

How can we make that question if there's a lot to learn about the brain? It is as complex as the universe.

1 year ago
Permalink

Comment has been collapsed.

A lot of very good stuff about current AI developments you can find in podcasts on https://www.youtube.com/@lexfridman/videos. They are quite lenghty, but those are talks between top scientists in the field, which are still able to talk about this stuff in a very understandable way for an average viewer. For the beginning I would suggest 3 talks with Max Tegmark:
https://www.youtube.com/watch?v=Gi8LUnhP5yU
https://www.youtube.com/watch?v=RL4j4KPwNGM
https://www.youtube.com/watch?v=VcVfceTsD0A
Btw, Lex Fridman running the show is really incredible and there is a lot of different stuff to be found there... he at times talks with folks like Elon Musk and Gary Kasparov, but many of his guests, that I've never before heard about, are actually having even more interesting things to say. Generally, it's "let's use our scientific knowledge to guess, where is the world going to" trip.

1 year ago*
Permalink

Comment has been collapsed.

They're masterworks all. You can't go wrong.

1 year ago
Permalink

Comment has been collapsed.

Sign in through Steam to add a comment.