Desirability of combining biological computing with silicon computing
S. B. Khadkikar, Department of Atmospheric and Space
Sciences,
Erach A. Irani, IKB Research Institute,
We are introducing a method to develop and fabricate a biological computer – silicon computer hybrid which can make decisions that no computer can at present and crunch data at the speeds of silicon computers. Fabricating the hybrid computer may require the skills of several technologists devoting part-time efforts. For the minimum, a biotechnologist is required who combines the skill of a microbiologist, electronic engineer, computer programmers and computer scientists at a minimum. In this speculative article, research is cited that shows how living cells have been joined to silicon computers. Also cited, is proof of “bacterial intelligence”. We are speculating that bacteria can be evolved using carbon nano-tubes that are semi-conductors to form a bio-electronic brain that can solve all the world’s problems. A speculation as to the development of the CONSCIOUS bio-electronic information processing bacterial unit (a CBEIPBU, pronounced “see-beep-boo”, like a silicon transistor in electronic computers) using nano-tubes is also given. How to make a CBEIPBU and how to identify a CBEIPBU in the bacteria is given.
The implication of
electronic speeds for the CBEIPBU and the bio-electronic brain is that what we
learn as humans in 40 years, the bio-electronic brain made out of CBEIPBUs will learn in seconds or minutes.
The species of
bacteria has been identified as the Geobacter
bacteria (www.geobacter.org) that
connect directly to electrodes and exchange electrons with an electrode. They join together and use nano-wires (or pili) and make biofilms on the electrodes. Since these Geobacter
bacteria respond to electricity, they will directly train with patterns such as
left-drift, right-drift, simple delay and later on XOR on a grid of
electrodes. Use of these Geobacter bacteria makes the task of demonstrating
trainable intelligence in bacteria trivial.
A solution containing
both Geobacter (to connect to electrodes by
exchanging electrons) and E.Coli (for its
intelligence since it lives in the human intestine and works with bacteriophages to exchange genes) can demonstrate even more
trainability and intelligence. Putting a
300 * 300 (or even 30 * 30) grid of electrodes in a solution and training
time-based left-drift, right-drift, associativity and
XOR should demonstrate bacterial programmability. Mutating these bacteria with carbon nanotubes should demonstrate even faster electronic speeds
as the bacteria form transistors with the carbon nanotubes
and mutate to evolve how to use them.
Use of “magnetotactic” bacteria along with “geobacter”
bacteria is also recommended.
We are aware of the work done in programming bacteria by transferring genes [14] and integrating our method of harnessing the consciousness of bacteria with this method should result in sensors for vision, etc. Integrating synthetic biology with this method of harnessing and evolving bacterial consciousness should also be useful.
Silicon computer programs are very good at data-crunching to come up with summaries of data. However, a human being is using the data. A human being is an example of a biological computer which has the “Observer” built into it. “Observer” is a notion from quantum mechanics. A human being is capable of making DECISIONS from what the computer reports. Computers are not capable of making DECISIONS. Computers are made of transistors made of non-living silicon. We introduce the notion of bacteria modified with nanotubes to work at electronic speeds. Nanotubes are semi-conductors and the bacteria should form CBEIPBUs or Conscious Bio-Electronic Information Processing Bacterial Units that are conscious. Many CBEIPBUs should network together to form a conscious computer.
Lots of strategies have been tried to make computers mimic biological systems. However, we contend that all these attempts will fail. Computers CANNOT MIMIC biological systems in their ability to make decisions. Computers CANNOT IMITATE biological systems.
The ability to make decisions is required in “self-programming” of computers, the ability to interview users and to write programs for a digital computer. Computers have become very good at transforming computer programs through compilation, however, to-date they are not able to make ANY DECISION. Computers are also very good at number-crunching and lightning fast data-processing, however since they lack CONSCIOUSNESS, they cannot make ANY DECISION.
2. What is Consciousness and what is a Decision?
Consciousness is intimately connected with the ability to make Decisions.
Consciousness requires an ability to somehow sense your Own Self in the Universe around you to the extent that you can perceive it and make transformations that allow your own Self to continue existence. We hypothesize that this Consciousness is innate in all living things. Bacteria are now hypothesized to contain Intelligence [2,3,4]. Bacteria work in bio-films and as bacterial consortiums to interact socially to achieve tasks. It takes intelligence for an organism or even perhaps a cell of a body to achieve its own tasks. A Decision is the ability to make the organism direct itself towards a goal. Computers lack consciousness and hence have no ability to recognize their environment or to have a goal. They are only good at data-processing.
25,000 rat neurons can fly an F22 simulator simulation
through difficult requirements[5]. These rat neurons learned to fly the F22
simulator in 15 minutes. [http://dsc.discovery.com/news/briefs/20041018/brain.html] Dr Eschel [4] argues
successfully for bacterial intelligence BUT HE DOES NOT USE TRAINING like we
propose using electronics to HARNESS and PROGRAM bacterial intelligence.
The authors have proposed how to make a collective identity of bacteria to enable the bacteria do tasks [1]. There is also an achievement of a researcher making a NATURAL urinary-bladder by growing cells on a scaffolding which dissolved away [7,8]. So, perhaps, it is not so difficult to make a brain out of neurons but which interfaces to a computer although the brain is a complex organ. A slime mould has been connected to a robot to make the robot move. [6].
Synthetic biology is not harnessing the natural intelligence and consciousness of biology. Instead it is merely trying to design “biological circuitry” to do human-defined tasks. One of the authors, had studied for his PhD thesis proposal in 1988, a work by Robert Blum, a discovery system, RX, which wished to use a computer based discovery system that would self-operate in a loop to obtain new facts. Such a system has yet to come in existence, because, we speculate, that no digital computer has the ability to make decisions. Biological systems have life, with life comes the ability to be conscious, have a recognition of one’s environment, and make decisions.
4. What is the future of computing?
The future of computing is a biological living computer joined to a silicon or non-living computer. The biological computer makes decisions and the silicon computer crunches data rapidly. We speculate that training biology to make decisions should be easy, that is what Life is all about hopefully, advancing consciousness and thus being able to recognize and manipulate more of the Universe.
For an article on consciousness, one can read http://home.att.net/~ag2kh/consciob.htm
Consciousness is not easy to define, perhaps it can be defined as what a human being or living being has and what a computer, although imitating intelligence and far superior to most people in chess-playing, does not have. A chess-playing computer is not aware it is playing chess, a human being is very much aware that he/she is playing chess.
By all these amazing feats of biology, one can indicate, that biology is far more self-repairing than computers and the flexibility of biology to make decisions when connected to a computers data-processing capabilities, should hopefully yield computers capable of understanding the human body.
5. FABRICATING A BIO-BRAIN (
In the 1960’s, there was a lot of work done on artificial intelligence by psychologists specifying how the brain thinks. Similarly, it is useful to speculate how a bio-brain can be built out of bacteria incorporating nano-particles. Speculation at least provides a platform for discussion that leads to better experimentation. Making a human brain out of neurons is perhaps too complicated, and one does not wish to duplicate the human brain, one wants a biological component connected intimately to a computer system.
In order to fabricate a bio-brain, one grows a colony of bacteria on a 3-dimensional grid. The 3-dimensional grid has surfaces on which bacteria can grow, ‘food holes’ from which food can be pumped using MEMS (Micro electronic mechanical structures). Alternatively, we can use a grid of fine plastic pipes with holes in them and use pumps to pump the food. The food can be a mixture of proteins and carbohydrates at different locations on the grid, or a mixture of proteins and carbohydrates uniformly throughout the grid. An electric grid which serves to provide electric current used for heating up points on the grid, and/or lighting up diodes, and/or generating magnetic waves which may influence the bacteria. We may use for example magnetosomes bacteria or bacteria with carbon nano-tubes incorporated in them.
The first task is to build up an associativity between food and electric impulses that generate heat, light, or magnetic waves. The grid supplies food through micro-plastic pipes. The grid also communicates information about where the food is supplied and in what quantity through intensity of electric currents. This should lead to an associativity between food and electric currents. If one speculates, that biology tends to increase its intelligence in the presence of the right intelligent environment, then one speculates that the associativity will be formed.
If an associativity is formed, then one impresses different patterns on the grid to increase learning. In “neural networks” [9] branch of computer science/artificial intelligence a neural network capable of associativity learned to simulate language making the same type of mistakes that children do when learning speech and do computations. Therefore, teaching associativity to a bacterial network is the foundation of intelligence in bacteria.
6. HOW TO INITIATE A FEEDBACK BETWEEN THE BACTERIAL NETWORK BRAIN AND THE COMPUTER SYSTEM
The input to the bacteria is the food pattern and the electric current pattern. Since the bacteria have nanotubes, we can have sensors on the 3D grid which detect the movement of the bacteria. These sensors detect the motion of the nanotubes and report back the movement of the bacteria to the computer system. The computer system then responds by giving food and electric flow to the bacteria. Only reward is used in training, since punishment is automatic in nature as lack of food and electric flow.
7. MAKING OF DIFFERENT SENSOR REGIONS IN THE BACTERIAL NETWORK BRAIN
The human brain has different regions. Assuming a associative bacterial network brain is formed we concentrate the electric current and food for different signals in different parts of the bacterial network brain to concentrate the bacteria in different regions.
8. CONNECTIONS BETWEEN DIFFERENT REGIONS OF BACTERIA
Biological nerves are slow compared to electronic circuits. By using the computer to conduct information between different bacterial regions, we supply bacteria with a rapid path for exchange of information.
9. COMPARISON BETWEEN NEURONS AND BACTERIAL COMPONENT
Neurons are slow compared to the bacteria which have semi-conducting properties built into them from the very beginning. Further the bacteria exchange information using electronic circuits. Thus they should be much faster than any biological brain.
10. IS THERE ANY WAY BESIDES MAKING A BACTERIAL BRAIN TO CONNECT TO COMPUTERS?
Advantages of designing a brain using bacteria and semi-conducting nano-tubes from the ground up is that we are designing a very fast conscious brain. Bacteria are among the smallest units of life with extremely fast mutating characteristics and hence they are chosen. Bacteria mutate at the rate of millions of generations a day under adverse conditions. Genes of neurons can be placed in some bacteria so that the bacteria may incorporate the genes.
Using neurons (human or animal) will not result in an extremely fast circuit that can be integrated with a computer and yet be the size of a football stadium if necessary. One needs a fast and large biological brain to work with all the knowledge on the internet and which can still suggest more experiments to deal with issues like decoding the human body’s disease patterns, perhaps even aging and death, alternative energy, quantum physics, and perhaps even time travel. In future, there may be limitations on the size of a brain built using neurons, a bacterial brain with nano-particles, and electronic computer-connected wires as nerves should have no size limitations.
Silicon has yet to make a single decision. As a computer programmer, I point out the paradox that a computer can now play better chess than most people, but it does not know that it is playing chess. Examples of electronic computers that cannot make a decision and hence are not trusted are the expert system MYCIN, the expert system INTERNIST, and most famously even the search engine Google which is incapable of differentiating what content should be shown to children and adults. Google is also not capable of deciding why a search is being done when a search is done nor can it do a search refined on all the information it stores on a person using Gmail. Google is also not capable of imparting knowledge or even formulating search criteria in a question-answer fashion.
10. Is evolving
bacteria with nanoparticles an original idea?
The seeds of this idea originated in a talk given at Pune University. Further refinements were done and a US Patent was filed on the concept as US Patent Application number 20060024810 with a filing date of July 27, 2004. Jocelyn Paine, the editor of Dr. Dobbs AI Newsletter published the idea with some very good instances of biological intelligence [9]. Some instances where biology has been connected to electronics to harness biological intelligence have been noted.
11. Circulatory
channels with liquid flow
There can be circulatory channels in the grid with liquid flow to remove waste-products and perhaps even to carry nutritious liquids throughout the grid. The food-holes can put the nutritious liquids inside the circulatory channels.
12. Ethics of killing
thinking bacteria and how to resolve them
If the bacteria in
the grid are proven to have a collective identity that can think, then one does
not want to kill them to make way for a new brain. One just grows the grid and let the newer
thinking bacteria co-operate with the older thinking bacteria. This should resolve the ethics of killing
thinking bacteria.
12. Implication of bio-electronic speeds for the
bio-electronic brain, or why design the bio-electronic
neurons to be bio-electronic from the ground up.
Rat neurons as used
in [5] will only give us a brain as fast as a human brain. We need electronic speeds to be incorporated
into the living cell from the ground up.
Electronic speeds are needed so that the bio-electronic brain can learn
rapidly at electronic speeds. What we
learn in 40 years it should learn in minutes or even seconds. Only then can the bio-electronic brain learn
all the knowledge of the internet and decode the human body’s biology and
suggest cures in the next few years, instead of decades.
13. Fabrication of the bacterial bio-electronic
information processing unit, the equivalent of the silicon transistor.
It has been stated
on the internet that carbon nano-tubes penetrate
bacteria and go straight to the nucleus, killing the bacteria. Now, one speculates, why do the carbon nano-tubes go straight to the nucleus? Is that a random occurrence or just a
biological happening or are bacteria consciously inviting the nano-tube into the nucleus so that the bacterial nucleus
can harness the information-processing abilities that result from using nanotubes in their genetic computational operations.
Now, the first task
before the researcher is to fabricate the bio-electronic information processing
unit (CBEIPBU) inside a bacteria’s nucleus.
The good thing is that the researcher is not responsible for fabricating
a CBEIPBU, but the bacteria will fabricate the CBEIPBU out of millions of
mutations once it is exposed to nano-tubes of
different sizes and characteristics. The
speculation is that the CBEIPBU will be formed because the bacteria that
succeeds in processing information electronically as well as chemically, has an
advantage over other bacteria and hence dominates. Once a usable CBEIPBU is formed inside the bacteria,
it will share the genes for the CBEIPBU with its neighboring bacteria and the
CBEIPBU will dominate and we will be able to see all the bacteria with the
carbon nanotubes used inside the CBEIPBU. Of course, one may expect that a particular
form (single-walled SWT or multiple-walled) and size of carbon nanotube will be used preferentially but even that is not
necessary as biology is flexible.
Even if a CBEIPBU is
not formed out of carbon nanotubes, one has to
experiment with different “things” that evolve bacteria to accelerate bacterial
information processing using electronics, until a CBEIPBU is formed in
bacteria.
14. Training once the Conscious bio-electronic
information processing unit (CBEIPBU) is formed in bacteria.
Once the CBEIPBU is
formed the bacteria should mutate or think very rapidly in response to
challenges it is faced with. It will not
be thinking at chemical speeds, it will be thinking at a speed between chemical
speeds and electronic speeds. Problems
of associativity it should solve rapidly.
15. Which bacteria species to use ?
This is best decided
by a microbiologist. But a species of
bacteria “geobacter” forms biofilms
and passes electrons via nano-wires called pili and is used in fuel cells. Perhaps this species can be used initially
[11].
More information on
the Geobacter bacteria can be found on www.geobacter.org. [13]. The Geobacter bacteria connect directly to electrodes and
exchange electrons with an electrode.
They join together and use nano-wires (or pili) [12] and make biofilms on
the electrodes. Since these Geobacter bacteria respond to electricity, they will
directly train with patterns such as simple delay and later on XOR on
electrodes. This eliminates the
previously mentioned tedious procedures of training the bacteria using
“food-holes” and MEMS and pumps and mutation using carbon nano-tubes
although this background could be used in future.
16. Combination of bacterial species to use and simple
experiments to do.
A solution containing both Geobacter (to connect to electrodes by exchanging electrons) and E.Coli (for its intelligence since it lives in the human intestine and works with bacteriophages to exchange genes) can demonstrate even more trainability and intelligence. Putting a 300 * 300 (or even 30 * 30 initially) grid of electrodes in a solution and training left-drift, right-drift, time-based associativity and XOR should demonstrate bacterial programmability. Mutating these bacteria with carbon nanotubes that are semi-conductors should demonstrate even faster electronic speeds once the carbon nanotubes make transistors with bacterial outer-membrane proteins.
17. Tumbling
behavior of bacteria towards food. Distributed bio-intelligence versus Central Processing Unit of
computers.
I
believe the statement below of bacterial motion shows that bacteria have a
distributed form of intelligence. Is
there any reason why bacteria which have developed a 50 protein flagellum
“motor” in which all proteins have to be present for the flagellum “motor” to
work cannot be trained. Geobacter
bacteria have protein insulated 5-10 nano meter wide “pili” or nano-wires that are 10 micrometers long and conduct
electricity across the length of the pili. Is it reasonable that the geobacter
bacteria will respond to electric patterns since they exchange electrons with
their environment.
Bacteria may not have a CPU or central processing unit as in computers
but even the human brain does distributed processing where memory and thinking
is integrated in the neurons.
Central brains are not
required for activity in biology.
Despite their name, jellyfish are actually not fish! Jellyfish are made
up of over 95% water, and they do not have brains, hearts, gills, bones, or
blood. (http://www.edhelper.com/AnimalReadingComprehension_42_1.html)
Bacteria move towards food, for eg. by "blundering" towards it. Only they
"happen' to make fewer blunders when they are moving towards it than when
away from it.....it's called 'tumbling" motility. bacteria randomly tumble after swimming short distances and
after the tumble they move in a random direction and again tumble and so on....
but, when moving towards food their freq. of tumbling decreases and hence they
"end up" moving "towards" it.
18. Potentially cheap method of making carbon nanotubes
A potentially cheap
method of making carbon nanotubes is given in
[10]. In this article, new method
for making multiwalled carbon nanotubes
by heating grass in the presence of oxygen has been demonstrated. The nanotubes were about 1 micron long and 30 to 50 nm in
diameter.
19. Miniaturization
of bacterial computing elements by the bacteria themselves
If bacterial computing (or mutation
computing) succeeds, then each individual bacterium should develop several
small computing elements (like pili or even
smaller). A bacterium is of the order of
a micrometer, the individual computing elements will be much smaller. The pili in
geobacter are only 3-5 nanometers in
width and upto 10 micrometers in length. ( http://www.geobacter.org/research/nanowires). There are more characteristics of pili of Geobacter
such that the proteins that coat it are non-conducting too (to make the pilin like a wire to conduct electricity) (letter to Nature) [16].
20. Length of the
genome code in bacteria and computing requirements
As biofilms, the bacteria can compute in networks. We are also trying to make the bacteria
evolve using carbon nanotubes as tools. So we expect the length of their genome code
to go up. The length of the genome code
should not be a problem for the computing requirements if one considers that
bacteria compute socially. The human
genome code is not the longest in the animal kingdom, yet the amount of
information on the internet is immense.
Also, the bacteria
have evolved a 50 protein driven motor for their flagellum. The motor is irreducible,
it requires all 50 proteins to function.
If bacteria
can do that, geobacter can certainly sense left-flow,
right-flow, and other patterns of current especially since they can be rewarded
chemically.
21. Use of magnetotactic bacteria along with geobacter bacteria.
It would be good to
promote competition and co-operation between two species of bacteria. Here the magnetotactic
bacteria come to mind. They can be moved
by a magnetic field from one electrode to another during the training sequence
of the geobacter bacteria. The geobacter
bacteria might detect the magnetotactic bacteria
moving and might move along with them.
Thus a combination of magnetotactic bacteria
and geobacter bacteria might be the best combination
(instead of E. Coli).
This is especially inspired by the article “Research Finds Magnetic
Bacteria Misfits”
( http://dsc.discovery.com/news/briefs/20060123/magneticbacteria_pla.html
).
22. Why carbon nanotubes are specifically chosen as tools?
Carbon
nanotubes are made of carbon which may integrate well
with proteins trying to bind around them.
Carbon nanotubes have good strength
characteristics which gives the bacterium that adapts to it a good claw to
defend itself against bacteria with and attack other bacteria for food. Carbon nanotubes
enable the bacteria to digest harder food and even perhaps cancerous cells (a
possible cancer treatment). SWT (single-walled
tubes) have magnetic properties and electrical semi-conducting properties which
are good when forming transistors.
23. What can the
twin combination of carbon nanotubes and electronic
training result in?
Carbon nanotubes
constitute wealth for the bacteria once the bacteria adapt to them. This should induce loose collectives and
induce specialization in the bacteria and differentiation of function. The bacteria may form balls where the outer
bacteria adapt to defense against carbon nanotubes
and the inner bacteria nourish the outer bacteria. The bacteria that can integrate electronic
thinking by using Carbon nanotubes as transistors
think faster than the others and help may mutate/evolve to do so under a electronic training pattern. Initially, geobacter
bacteria may switch around current and later on learn to think
electronically.
The twin impetus of electronic training and carbon nanotubes can cause the genome length of the bacteria to go
up.
24. Number of transistors in an electronic CPU and number
of bacteria in a colony
An electronic Intel
dual-core CPU has ONLY 230 million transistors and this is considered a large
number. (http://www.trustedreviews.com/article.aspx?page=2753&head=0). A bacterial colony has 10^9 to 10^10 or is it
10^12 bacteria (Ben Eschel). We can make different small colonies
connected by electronic circuits grids or one colony
connected by numerous grids and communicating with each other. Also, each geobacter,
even if it has only 10 pili and 100 nano-tube transistors, and a “native intelligence factor
equivalent to 100 transistors” will give you 10^10 (or is it 10^12) * 100
native transistors giving you a circuit of 10^14 computing elements. So an Intel CPU by
Will have to go by
10^ 14 / (250 * 10^6) = say
4 * 10^5 factor increase which will take at log ( 4 * 10^5 ) /
log 2 = (log 4 + 5 ) / .3010 = 5.6 / .3010 = 18.645 * 1.5 years = say
27.9 years. Of course, the speeds of the CPU will be
fantastic. So we assume that the
intelligence factor is (CPU-speed squared * number of transistors) = (1.5 * 1.5
* 1.5) =
3.375 or even say log (8)
Therefore we
calculate log ( 4 * 10^5) / log 8 = 6.203 * 1.5 years
= 9 years.
Therefore an INTEL
CPU (silicon cpu) should
take 9 years to be hardware-wise as capable as a colony of bacteria.
25. CONCLUSION AND FUTURE OUTLOOK
The integration of microbiology with computer technology should result in a hybrid computer which can take decisions rapidly and compute data rapidly. This should enable all the information on the internet to be integrated to suggest new experiments to solve humanity’s requirements. Training of this complex bio-silicon composite brain will be the next step in creating a decision making and rapid computing system. Training this bio-silicon brain should be feasible and simpler than making a pure silicon brain neural network which cannot be made functional because decision making is a consciousness based activity and consciousness is a property of life.
Bibliography
(www.google.com search for “bacterial intelligence”)
(www.google.com search for “bacterial intelligence”)
(www.google.com search for “bacterial intelligence Eschel”)
5.http://iwarrior.uwaterloo.ca/?module=displaystory&story_id=1652&format=html&edition_id=35
(www.google.com search for “25,000 rat neurons flying F22 simulator”)
(www.google.com search on making a robot move using a slime cell)
8. “First Bladders Grown in Lab Transplanted,
Breakthrough Shows Promise for Creating Other Human Organs”,
http://www.washingtonpost.com/wp-dyn/content/article/2006/04/03/AR2006040301387.html
[Nanotechnology 16 1192] (6.13.05)
11. “Scientist Revs Up Power of Microbial Fuel Cells in Unexpected Ways”,
http://www.oocities.org/erach27/BacterialBioFilmMakeElectricity.txt
12. http://www.geobacter.org/research/nanowires/
13. Information on Geobacter bacteria. http://www.geobacter.org
14. Christopher Voigt. Online internet lecture. Programming sight, touch, and thought into E.Coli. http://esmane.physics.lsa.umich.edu/wl/external/ICSB/2005/20051020-umwlap001-03-voigt/real/sld006.htm
(www.google.com search for “programming bacteria”)
15. http://www.oocities.org/erach27 for an index of papers in this area on the internet by Erach A Irani. Read the Godel’s theorem and Turing theorem article there.
16. Gemma Reguera1, Kevin D. McCarthy, Teena Mehta, Julie S. Nicoll, Mark T. Tuominen & Derek R. Lovley, letter to Nature Vol 435|23 June 2005, “Extracellular electron transfer via microbial nanowires”.
BIOGRAPHY
Prof. S. B. Khadkikar, MSc, PhD is an internationally recognized theoretical physicist and is a retired senior professor from Physics Research Laboratory, Ahmedabad. He is cited in Marquis Who’s Who (science and engineering, world, Asia) in the years 1996-2006. He has about 80 publications in renowned international journals, mainly in nuclear physics.
Dr Erach A. Irani, B.Tech Computer Science (IIT Mumbai, India), M.Tech Computer Science (University of Minnesota, Minneapolis, USA), PhD Computer Science (University of Minnesota, Minneapolis) has worked all his life in Computer Science and now theorizes with Prof S. B. Khadkikar. He is a US Citizen. He has over 20 publications to his credit while doing his studies at the University of Minnesota, mainly in computer science and applications of Computer Science to medicine.