#4393088 - 12/04/17 03:29 AM
Google's AI Built Its Own AI That Outperforms Any Made by Humans
|
Joined: Jun 2002
Posts: 11,946
Crane Hunter
Veteran
|
Veteran
Joined: Jun 2002
Posts: 11,946
Master Meme-er
|
Google's AI Built Its Own AI That Outperforms Any Made by Humans
In May 2017, researchers at Google Brainannounced the creation of AutoML, an artificial intelligence (AI) that's capable of generating its own AIs. More recently, they decided to present AutoML with its biggest challenge to date, and the AI that can build AI created a 'child' that outperformed all of its human-made counterparts. The Google researchers automated the design of machine learning models#using an approach called#reinforcement learning. AutoML acts as a controller neural network that develops a child AI network for a specific task. For this particular child AI, which the researchers called NASNet, the task was recognising objects - people, cars, traffic lights, handbags, backpacks, etc. - in a video in real-time. AutoML would evaluate NASNet's performance and use that information to improve its child AI, repeating the process thousands of times. When tested on the ImageNet image classification and COCO object detection data sets, which the Google researchers call "two of the most respected large-scale academic data sets in computer vision," NASNet outperformed all other computer vision systems. According to the researchers, NASNet was 82.7 percent accurate at predicting images on ImageNet's validation set. This is 1.2 percent better than any previously published results, and the system is also 4 percent more efficient, with a 43.1 percent mean Average Precision (mAP). Additionally, a less computationally demanding version of NASNet outperformed the best similarly sized models for mobile platforms by 3.1 percent. A view of the future Machine learning is what gives many AI systems their ability to perform specific tasks. Although the concept behind it is fairly simple - an algorithm learns by being fed a tonne of data - the process requires a huge amount of time and effort. By automating the process of creating accurate, efficient AI systems, an AI that can build AI takes on the brunt of that work. Ultimately, that means AutoML could open up the field of machine learning and AI to non-experts. As for NASNet specifically, accurate, efficient computer vision algorithms are highly sought after due to the number of potential applications. They could be used to create sophisticated, AI-powered robots or to help visually impaired people regain sight, as one researcher suggested. They could also help designers improve self-driving vehicle technologies. The faster an autonomous vehicle can recognise objects in its path, the faster it can react to them, thereby increasing the safety of such vehicles. The Google researchers acknowledge that NASNet could prove useful for a wide range of applications and have open-sourced the AI for#inference on image classification and object detection. "We hope that the larger machine learning community will be able to build on these models to address multitudes of computer vision problems we have not yet imagined," they wrote in their blog post. Though the applications for NASNet and AutoML are plentiful, the creation of an AI that can build AI does#raise some concerns. For instance, what's to prevent the parent from passing down unwanted biases to its child? What if AutoML creates systems so fast that society can't keep up? It's not very difficult to see how NASNet could be employed in automated surveillance systems in the near future, perhaps sooner than regulations could be put in place to control such systems. Thankfully, world leaders are working fast to ensure such systems don't lead to any sort of dystopian future. Amazon, Facebook, Apple, and several others are all members of the Partnership on AI to Benefit People and Society, an organisation focused on the responsible development of AI. The Institute of Electrical and Electronics Engineers (IEE) has proposed ethical standards for AI, and DeepMind, a research company owned by Google's parent company Alphabet, recently announced the creation of group focused on the#moral and ethical implications of AI. Various governments are also working on regulations to prevent the use of AI for dangerous purposes, such as autonomous weapons, and so long as humans maintain control of the overall direction of AI development, the benefits of having an AI that can build AI should far outweigh any potential pitfalls. http://www.sciencealert.com/google-s-ai-built-it-s-own-ai-that-outperforms-any-made-by-humansinb4 the singularity
Yeah so I just hope I'm able to play a few absolutely perfect AI create games before eventual death by snu-snu at the hands of some 9000 IQ sexbot, plz.
|
|
#4393090 - 12/04/17 03:39 AM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: Crane Hunter]
|
Joined: Mar 2006
Posts: 2,450
vocatx
Member
|
Member
Joined: Mar 2006
Posts: 2,450
Voca, Texas
|
Considering the wife and I are on season 5 of 'Person of Interest', I read this with more than a little trepidation...
4H_V-man
System Specs: MSI 870 A-G54 AMD X4 Phenom II @ 3.4 ghz 8 Gb DDR3 1333 RAM 1 Tb Western Digital 250 Gb Western Digital 320 Gb Seagate Radeon HD 6970 graphics Thermaltake 650 Modualar power supply Thermaltake Element V full tower
|
|
#4393140 - 12/04/17 11:43 AM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: Crane Hunter]
|
Joined: Apr 2001
Posts: 121,489
PanzerMeyer
Pro-Consul of Florida
|
Pro-Consul of Florida
King Crimson - SimHQ's Top Poster
Joined: Apr 2001
Posts: 121,489
Miami, FL USA
|
Amazon, Facebook, Apple, and several others are all members of the Partnership on AI to Benefit People and Society, an organisation focused on the responsible development of AI.
The Institute of Electrical and Electronics Engineers (IEE) has proposed ethical standards for AI, and DeepMind, a research company owned by Google's parent company Alphabet, recently announced the creation of group focused on the#moral and ethical implications of AI.
Uh huh... We all know it wouldn't be too long before some "rogue" country or terrorist group would get its hands on the code. Nothing is 100% hack proof.
Last edited by PanzerMeyer; 12/04/17 11:43 AM.
“Whoever fights monsters should see to it that in the process he does not become a monster. And if you gaze long enough into an abyss, the abyss will gaze back into you.”
|
|
#4393163 - 12/04/17 02:25 PM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: Crane Hunter]
|
Joined: Dec 2000
Posts: 13,225
No105_Archie
|
Veteran
Joined: Dec 2000
Posts: 13,225
N. Atlantic east of Canada
|
DeepMind, a research company owned by Google's parent company Alphabet, recently announced the creation of group focused on the#moral and ethical implications of AI. that sounds like the intro to a new Terminator style movie !!!
Archie Smythe
carpe diem
|
|
#4393170 - 12/04/17 03:14 PM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: Crane Hunter]
|
Joined: Apr 2002
Posts: 5,503
Pooch
Hotshot
|
Hotshot
Joined: Apr 2002
Posts: 5,503
Orlando, FL
|
I don't think they are, but you have to walk before you can run. It's a move towards that, I think.
"From our orbital vantage point, we observe an earth without borders, full of peace, beauty and magnificence, and we pray that humanity as a whole can imagine a borderless world as we see it, and strive to live as one in peace." Astronaut William C. McCool RIP, January 29, 2003 - Space Shuttle Columbia
|
|
#4393172 - 12/04/17 03:28 PM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: Pooch]
|
Joined: Apr 2001
Posts: 121,489
PanzerMeyer
Pro-Consul of Florida
|
Pro-Consul of Florida
King Crimson - SimHQ's Top Poster
Joined: Apr 2001
Posts: 121,489
Miami, FL USA
|
I don't think they are, but you have to walk before you can run. It's a move towards that, I think. If fully self-aware and abstract thinking AI ever does happen, it won't be during my lifetime. Personally, I don't see that ever happening. It would be like giving human intelligence to a dog or a horse.
Last edited by PanzerMeyer; 12/04/17 03:28 PM.
“Whoever fights monsters should see to it that in the process he does not become a monster. And if you gaze long enough into an abyss, the abyss will gaze back into you.”
|
|
#4393194 - 12/04/17 05:00 PM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: PanzerMeyer]
|
Joined: Dec 2008
Posts: 26,566
wheelsup_cavu
Lifer
|
Lifer
Joined: Dec 2008
Posts: 26,566
Corona, California
|
Amazon, Facebook, Apple, and several others are all members of the Partnership on AI to Benefit People and Society, an organisation focused on the responsible development of AI.
The Institute of Electrical and Electronics Engineers (IEE) has proposed ethical standards for AI, and DeepMind, a research company owned by Google's parent company Alphabet, recently announced the creation of group focused on the#moral and ethical implications of AI.
Uh huh... We all know it wouldn't be too long before some "rogue" country or terrorist group would get its hands on the code. Nothing is 100% hack proof. Google's damn password recovery loop may as well be... Wheels
|
|
#4393196 - 12/04/17 05:12 PM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: Crane Hunter]
|
Joined: Jan 2006
Posts: 9,624
Mr_Blastman
Hotshot
|
Hotshot
Joined: Jan 2006
Posts: 9,624
Atlanta, GA
|
Panzer: What is consciousness?
For a human, it is awareness of your surroundings and the ability to use that awareness to adapt and change.
In your brain, what allows you to have consciousness, assuming our bodies are nothing more than machines? I wager our adaptive neurons and synapses, how they can form new branches with different tranches, creating webs within webs and the ability for each and every neuron to process data independently from one another while simultaneously aware of their existing state.
If AI can adapt and change it is halfway to being self aware, and as it builds more and more algorithms independently of its source code, it is creating consciousness as it goes. The singularity is inevitable. AI will likely achieve consciousness at some point in the next one hundred years unless we install measures to prevent it, and even if we put those measures into place, AI may still figure out a way around them and outsmart us in the end.
So what to do? Limit AI's access to the real world. Air gap AI from machinery or networks that access physical devices is one place to start. But... once AI is installed in independently autonomous mobile platforms(aka robots), that grows far trickier.
|
|
#4393203 - 12/04/17 05:35 PM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: Mr_Blastman]
|
Joined: Jul 2002
Posts: 3,506
DM
Senior Member
|
Senior Member
Joined: Jul 2002
Posts: 3,506
Prague
|
Panzer: What is consciousness?
For a human, it is awareness of your surroundings and the ability to use that awareness to adapt and change.
In your brain, what allows you to have consciousness, assuming our bodies are nothing more than machines? I wager our adaptive neurons and synapses, how they can form new branches with different tranches, creating webs within webs and the ability for each and every neuron to process data independently from one another while simultaneously aware of their existing state.
If AI can adapt and change it is halfway to being self aware, and as it builds more and more algorithms independently of its source code, it is creating consciousness as it goes. The singularity is inevitable. AI will likely achieve consciousness at some point in the next one hundred years unless we install measures to prevent it, and even if we put those measures into place, AI may still figure out a way around them and outsmart us in the end.
So what to do? Limit AI's access to the real world. Air gap AI from machinery or networks that access physical devices is one place to start. But... once AI is installed in independently autonomous mobile platforms(aka robots), that grows far trickier. The thing is - the only entity you can be sure is conscious is yourself. All other entities could be emulating it via learned behaviors. Kind of like how some sociopaths appear charismatic. I have known plenty of people who wouldn't pass the Turing test But I feel it's dangerous to compare AI consciousness with human consciousness. It's developed/is developing in an alien way compared to how our consciousness came about. You might have to start about thinking how you could emulate a childhood with no idea of its true nature. It's said that if a lion could speak English that you still would have no idea about what it was talking about. And that's like only a couple of percent away DNA-wise.
"They might look the same, but they don't taste the same."
|
|
#4393208 - 12/04/17 06:09 PM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: Crane Hunter]
|
Joined: Jan 2006
Posts: 9,624
Mr_Blastman
Hotshot
|
Hotshot
Joined: Jan 2006
Posts: 9,624
Atlanta, GA
|
Good point DM. So are we conscious? Is our reality... real? What is reality, for that matter? Are you real? Or am I? Or are we both a figment of someone else's imagination? Perhaps we're old thoughts, and AI is new thoughts... Then we're simply passing leafs in the strengthening wind, lofted to the skies, never setting down again upon what was once our home. Destination yonder, terminal decay. Green matter fading darker shades of grey. Dust... without delay.
Intelligence is intelligence as consciousness is consciousness. Just because we can't understand AI consciousness doesn't mean that AI consciousness is not. Our "is this right or wrong," might be an AI's "is this efficient or inefficient." Zeroes and ones, sums of superpositions or anything in-between, there's no right way or wrong way, simply there is. So if an AI is capable enough to adapt, change and conquer, and does so faster than we can ever imagine, then it certainly is no longer unimaginable that an extraordinary AI is capable of extraordinary feats such as achieving consciousness through self-induced extreme evolution.
Last edited by Mr_Blastman; 12/04/17 06:13 PM.
|
|
#4393284 - 12/05/17 04:12 AM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: Crane Hunter]
|
Joined: Oct 2000
Posts: 6,842
Clydewinder
Mach 2 Infrared Orangutan
|
Mach 2 Infrared Orangutan
Hotshot
Joined: Oct 2000
Posts: 6,842
New Berlin, WI United States
|
I'm not worried. If AIs ever get too smart, we can just have Bohemia Interactive infect them with their Arma AI... next thing you know, all of the driverless vehicles will be swerving across 4 lanes back and forth hitting lampposts while the robot soldiers shoot each other in the back.
Robots are stealing my luggage.
|
|
#4393285 - 12/05/17 04:26 AM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: Clydewinder]
|
Joined: Jul 2002
Posts: 406
Franze
Member
|
Member
Joined: Jul 2002
Posts: 406
|
I'm not worried. If AIs ever get too smart, we can just have Bohemia Interactive infect them with their Arma AI... next thing you know, all of the driverless vehicles will be swerving across 4 lanes back and forth hitting lampposts while the robot soldiers shoot each other in the back.
The horror... The horror... An entire column of tanks, APCs, trucks, technicals, all stopped and lined up on a road... Waiting for that rabbit to move...
|
|
#4393293 - 12/05/17 08:57 AM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: PanzerMeyer]
|
Joined: Jun 2001
Posts: 3,646
Kodiak
Senior Member
|
Senior Member
Joined: Jun 2001
Posts: 3,646
Over the hills and far away.
|
If fully self-aware and abstract thinking AI ever does happen, it won't be during my lifetime.
Panzer, I wouldn't be so sure of that. The steps they are taking in the short time we have computers and AI are very impressive. I think it will be step by step, leap by leap and suddenly it is running and we cannot catch up anymore.
I Want To Believe
|
|
#4393315 - 12/05/17 02:43 PM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: Crane Hunter]
|
Joined: Mar 2001
Posts: 11,077
semmern
Veteran
|
Veteran
Joined: Mar 2001
Posts: 11,077
Oslo, Norway
|
Oh Deep Thought, tell us the answer to the great question of life, the universe and everything!
In all my years I've never seen the like. It has to be more than a hundred sea miles and he brings us up on his tail. That's seamanship, Mr. Pullings. My God, that's seamanship!
|
|
#4393319 - 12/05/17 02:51 PM
Re: Google's AI Built Its Own AI That Outperforms Any Made by Humans
[Re: Crane Hunter]
|
Joined: Apr 2001
Posts: 121,489
PanzerMeyer
Pro-Consul of Florida
|
Pro-Consul of Florida
King Crimson - SimHQ's Top Poster
Joined: Apr 2001
Posts: 121,489
Miami, FL USA
|
Get back to me when I can have a full blown conversation with an android which can fully demonstrate abstract thought and emotions free from any pre-determined programming.
Last edited by PanzerMeyer; 12/05/17 02:51 PM.
“Whoever fights monsters should see to it that in the process he does not become a monster. And if you gaze long enough into an abyss, the abyss will gaze back into you.”
|
|
|
|
|
|
|
|
|
|
|
|
Exodus
by RedOneAlpha. 04/18/24 05:46 PM
|
|