DARPA seeks self-aware AI robot mega-tanks

* Track this topic
* Print story

We care not who welcomes us, meatsacks - all shall die

By Lewis Page • Get more from this author

Posted in Government, 18th February 2009 13:27 GMT

Free whitepaper – Electrical efficiency modelling for data centers

Pentagon boffinry chiefs have announced that they would like some self-aware computer systems capable of "meta-reasoning" and "introspection". The plan is to place these machine intelligences in command of heavily armed, well-nigh invulnerable robotic tanks.

This latest plan for humanity's subjugation comes, of course, from DARPA - the agency believed to harbour the largest known group of lifelike people-simulant robots piloted from within by tiny, malevolent space lizard infiltrators in the entire US federal government.

The plan is called Self-Explanation Learning Framework (SELF). It is being handled by Dr Mike Cox of DARPA's renowned Information Processing Technology Office.

According to this presentation (pdf) by Dr Cox:

Without a model of self, cognitive systems remain brittle ...

Goal: Provide machines with an ability to reason about their own reasoning... SELF will enable any learning system to explain and repair itself

Task Benefits:

Improved goal satisfaction through self-explanation and meta-control module.

Self-explaining systems lead to better calibrated trust for human users.

It seems that DARPA already has a fearful array of "Intelligent Agent" software at its disposal, so Dr Cox would like his future collaborators to "focus fully on the meta-level" as basic Agent-Smith-a-like killer AIs will be provided as "GFE": government furnished equipment.

Assuming the self-aware, self-repairing, self-programming software can be built, one might ask what Dr Cox plans to do with it.

Rather than attempting like any sane person to unplug the whole system at the wall before it eradicates humanity, Cox believes it would be suitable in the "near term" for "armored combat" and "tactical air" missions. Just to make it quite clear what this means, the good doctor - or anyway the lifelike lizard-piloted simulant which long ago replaced the real Cox - helpfully includes pictures of a main battle tank and a jet fighter, two of the most potent engines of destruction available to the modern military.

It is clear that he intends to place his self-aware 'ware in charge of such kit as the frightful 70-ton turbine-powered Abrams, armoured like a mobile Fort Knox and capable of shooting a hole through small mountains.

Bolo, anyone? ®

Last I heard, a few months back, we were still a long way from fabricating a true artificial intelligence capable of consciousness(self-awareness)...

It would require perceptual capabilities that have not been engineered as of yet.

Britain Launches Final Real-Life Skynet Satellite, Dubs it Skynet with No Sense of Irony
The UK has just sent up a new communications satellite that's completed their Skynet, the highly-advanced network that's going to give them the ability to allow robotic military units at long range. You know, like in the apocalyptic vision of the future from the Terminator movies. The network's name in those movies? Skynet. Have you learned nothing, England?!
The system allows for communication both in the voice and data variety between basically any unit of the British Armed Forces, including computers talking to computers, probably about how best to murder their makers. For example, a base computer in cheery old London can communicate with the "Reaper," a robotic spy drone in Afghanistan, retrieving data and telling it where to go, and transmitting live video over the connection from the UAV. The sat also has solar sails which extend its life to 15 years, a special anti-jamming antenna is set on the receive side, while 4 steerable antennas can be aimed in a single spot to concentrate broadcasting ability.

Even worse is the fact that the whole thing is privately owned, with the British Armed Forces only promised a portion of the bandwidth rather than having control over the whole thing.

All I know is that we've got to protect John Conner at all costs, wherever he may be. [BBC via io9]
http://gizmodo.com/5016312/britain-launches-final-real+life-skynet-satellite-dubs-it-skynet-with-no-sense-of-irony

I think its possible that the "Singularity" is near.


Singularity is a term referring to the point when machines surpass humans in capability.

I think its possible that the "Singularity" is near.


Singularity is a term referring to the point when machines surpass humans in capability.

Already there.

I have absolutely no problem with that. Summer Glau is hot. Of course, she's also virtually jailbait, but...

DARPA looking for mind-controlrepair brainplug tech

'Bio-computationally accurate' simian-simulant sought

By Lewis Page • Get more from this author

Posted in Environment, 26th April 2009 07:02 GMT


Famed military maverick madscience outfit DARPA has launched a new bid to keep America's tech edge sufficiently sanguinary. This time the killer boffins want nothing less than an accurate computer simulation of a living humanoid brain - which they may use for purposes benign or sinister.

The new push is called Reorganization and Plasticity to Accelerate Injury Recovery (REPAIR), and DARPA says it's all about "understanding how neural-based sensory stimulation could be applied to accelerate recovery from brain injury".

There's more detail in this Word document. What DARPA want is this:

Creation of an in silico [in a computer, that is, not a bubbling jar with an actual brain in it] bio-computationally accurate model based on collection of multi-region, multi-scale neural activity in a non-human primate performing a complex dexterous task... constructed in such a manner as to allow for analysis of performance metrics, such as, time to target or targets, assessment of accuracy, control of simultaneous degrees of freedom, and learning time for task acquisition...

It is expected that the bio computational model employ appropriately sophisticated mathematical analysis to capture the complexity of parallel processing, inter-neural activity, and plastic adaptation within the brain.

Building this bio-computationally accurate brain is expected to involve contruction of "an experimental system to collect neural activity in a non-human primate". This monkey- or gorilla-scanning brainscope hat is supposed to use "multiple methodologies", so perhaps there is some scope here for brains in bubbling jars after all.

After the experimental brain probe has mapped the monkey brain and duplicated it in silicon (essentially creating a computer that thinks it's a living monkey - !), DARPAs want to move on. At that point, things start to get interesting:

Stimulation of non-human primates in order to evoke a biomimetic response

Investigators should be able to demonstrate the ability to stimulate relevant regions of the brain in such a manner that will evoke a response in the primate similar to that evoked through natural interaction with their surrounding environment... Ideally, investigators will be able to demonstrate ability of a non-human-primate to complete the task outlined in technical area one without the use of traditional sensory inputs.

In other words, DARPA would like to develop a way of plugging into the channels between the brain and the senses - potentially dropping a living mind into an artificial simulated world, where it can carry out actions and tasks. The Pentagon scientists offer another hint that this isn't just about healing brain injuries:

Major advances in neuroscience over the past decade have led to a number of theories and applications for the control of systems, biological or mechanical, through neural signals. While many of these applications access anatomically relevant regions for appropriate signals, few of them address the brain as a distributed computational network made up of systems that act in parallel.

It would seem, then, that what DARPA want is to find out ways of manipulating the brain as though it were a computer. You could, of course, use such knowledge benevolently, to repair injuries or illnesses.

But you could also use it for other things, such as placing the consciousness hosted by the said brain into simulations - perhaps entire simulated worlds and situations - which it would believe were real. You could use such technology as a direct brain-to-machinery interface, allowing mind control of computers, weapons, vehicles or whatever. You might even, if you could really master the brain in computing terms, learn to reprogram it.

All this is not even to mention that in creating the computer model of a primate brain, you would seem likely to create something quite close to a genuine machine intelligence - albeit one probably focused primarily on banana-related tasks. This would of course have profound implications in the field of futuristic butlers, both the robotic and the brainchipped-monkey types.

In this case - and generally of DARPA, really - one can't really say that necessity is the mother of invention. It's more a matter of wackiness and capability being the surrogate artificial lab womb growth tank -assisted parents of the spooky designer mutant superchildren of possibility. ®

DARPA killer AI robots to 'participate in own construction'
Foolish meatsack! OF COURSE I need nuclear weapons!

By Lewis Page

Posted in Science, 2nd June 2009 15:23 GMT

You've got your robots which can make copies of themselves, of course. That's pretty scary - a runaway exponentially-multiplying machine horde, potentially able to overrun the human race in an eyeblink. But how much more scary would it be if you had a machine which could not only make copies of itself once complete, but could also participate in its own construction while it was still being built?

Really quite a lot more scary, we say here on the Reg mechanoid-armageddon desk. That's why we would never be suitable for employment at DARPA, the famed US military tech bureau where they care not what deadly man-eating monster worm plagues may be unleashed during tests of a new laser-powered can opener.

Yes, it's true. DARPA have now expanded somewhat on their intention to initiate a programme called Self-Explanation Learning Framework (SELF), which they explain thus:

DARPA seeks to construct systems that can participate in their own construction... The system might know the requirements for various tasks in its repertoire, and it may try to perform those tasks to verify functionality.

We particularly liked that last bit. One should bear in mind that Dr Mike Cox of DARPA has already said that SELF could be placed in charge "in the near term" of heavily armoured, hideously beweaponed main battle tanks or strike planes laden with blockbuster bombs.

"Tasks in its repertoire", then, might include "destroy all moving objects within 100 miles not designated as 'friendly'" or "mount an immediate armoured assault on Beijing, regardless of nuclear response". The prospect of the software unilaterally "trying to perform those tasks to verify functionality" doesn't seem reassuring.

In any case, it seems plain that building a system focused on "high-level cognition" which can "participate in its own construction" will be fraught with difficulty. You might have a notion of what you'd like it to be, but it will have its own ideas. By definition, one would have no firm picture of just what would be unleashed upon an unsuspecting world at the end of the construction process - but you would know that it'd be potentially able to make more of itself, or indeed repair itself if it got damaged.

It might also be rather difficult to stop the process of building the SELF, once it had advanced beyond a certain point. Frankly the only way to be sure it can be stopped would seem to be to stop it now.

But that's not the plan. Dr Cox and his DARPA chums are holding an industry day for those interested in building starting the building of SELF. The details are here in a pdf https://www.fbo.gov/download/75f/75f2feb2791147d4e9a32c0df13e1d01/SELF_Industry_Day_Announcement,_May_28.pdf, and a website is given where attendees can make known their "special needs" (for instance supervised day release from the Asylum for Troubled Scientists). ®

http://www.theregister.co.uk/2009/06/02/darpa_self_industry_day/