Pages

May 19, 2008

Limits to AGI, brain computer interfaces, nanomedicine and nanorobots

Some people scoff at mind uploading, human level or greater artificial intelligence or nanorobots that are able repair all damage to the human body

Here is a review of the current state of brain computer interfaces, brain simulation, and nanomedicine related medicine (cellular repair and rejuvenation) and science. I do not see anything stopping zettaflop level computers (one million times more powerful than the current best supercomputers). Zettaflop computers will enable human level brain simulations or greater. Brain computer interfaces already exist in many forms and are able to restore and create memories. There is nanoscale technology for precise delivery of drugs, genes and imaging agents into the cells of the human body and interaction with cells. Adult cells have been rejuvenated (made younger) by manipulating the cellular mechanisms. It will be a lot of hard work that will need to be funded to achieve the increased levels of capabilities, but it appears that there is a path to those levels and that it is not unreasonable to expect those levels to be achieved.

From the April IEEE Spectrum, an article on recent Brain-computer interface work which indicated large funding from DARPA for significant brain-computer interface progress by 2009.

2007 work for integrating an artificial 12000 neuron memory device with the human brain

Clearly those who predict that artificial brains will never reach humans levels feel that progress on hardware brain simulation or software brain simulation will reach limits or will not be integrated with more advanced brain/computer interfaces

Millions of neurons now and billions projected for several funded and active projects by around 2015

Will exaflop computers not be achieved ?


The Zettaflop design work won't pan out ?

10,000 neurons and 30 million synapses using a 22.8 teraflop supercomputer now. (45 times more power in a petaflop machine; 45,000 times for an exaflop, 45 million times for a zettaflop).

The researchers believe that the hardware for full brain simulation will be available in 2017.

A whole human brain has 100 billion neurons and 100 trillion synapses.

So what are the hardware, bandwidth or interface issues ?

I can also put together the current state of nanomedicine. With the ability of nanoscale devices to be targetted in the body at tumors or for delivery of gene therapy or delivery of sensors or imaging agents or drugs.

There is scientific/medical journal discussion on invivo rejuvenation of cells

There has been the analysis of the process to turn existing cells into rejuvenated cells. (A sequence of gene manipulations.)

So it appears that nanomedicine does not need to manipulate every molecule in the body. It merely has to extend and enhance existing alterations of processes in the body.

Which goes to the strategy of surveying what is working best now and enhancing and extending those methods instead of finding far more difficult plans and showing how those would have problems with physical limits.

Someone can show - look how little we understand about gravity and how hard or impossible anti-gravity is. OR someone can look at more efficient flight and space launch systems or look at enhancing magnetism for a ground based launch against the earth's magnetic field. There are many different options and strategies for any goal. Determine if the goals are worthwhile and find the best approaches for getting to worthwhile goals.

Flying cars are technically achievable, but 1.2 million deaths from regular ground based cars. Are flying cars a worthwhile goal ?
A technologically and economically practical flying car. but safety ? Insurance ?

FURTHER READING
New implantable device can extract stem cells from blood

Bloodstream robots and other technology from Engines of Creation

Enabling regeneration in humans

Progress to nerve and paralysis repair

Carbon nanopipettes for cellular surgery of organelles within the cell

Magnetically assembled nanotube tipped probes

0 comments: