9
$\begingroup$

Would a computer using biological components as an interface be any good? Would that protect it against hacking? Or provide more complex processing?

The way I'm going with this is that computers are at risk of getting hacked when they connect to the network. However as I understand biological computers AKA brains cannot be hacked like computers. So would using neurons to transmit and translate information between the processor and the transmitter/receiver get rid of the risk of a virus getting through? Sure neurons are much slower in terms of computational speed and may not have time to recover between actions, but would that interface provide any protection? If it doesn't then what would be a good alternative?

Also can using biological components provide some other advantage to a computer, like flexibility or better pattern recognition? Or would it be too much of a nuisance to be any help?

Edit: Thank you for your answers. This has been very informative.

$\endgroup$
  • 2
    $\begingroup$ Brains can not be hacked only because no one really tried hard. Unix and iOS at some time were also thought to be unhackable. $\endgroup$ – Alexander Dec 30 '19 at 20:06
  • 16
    $\begingroup$ @Alexander, people have too "tried hard"... and succeeded. Social engineering and hypnotism are just two of the more "mainstream" ways of manipulating brains, and that's without adding pharmaceuticals to the mix. There are people that specialize in ability to manipulate other people. $\endgroup$ – Matthew Dec 30 '19 at 20:10
  • 3
    $\begingroup$ "What would be a good alternative?" There is ample literature about the security of automated data processing systems. There is even a dedicated Stack Exchange site! (And I cannot believe that you believe that computer systems are vulnerable because their data connections are made of glass or copper instead of proteins. What importance can the material possibly have?) $\endgroup$ – AlexP Dec 30 '19 at 22:18
  • 4
    $\begingroup$ In addition to social engineering, there are plenty of examples of parasites that hack into the brains of their hosts and change their behavior to better suit the attacker. $\endgroup$ – Erik Dec 31 '19 at 10:53
  • 4
    $\begingroup$ Your understanding is incorrect. People's brains get hacked all the time. Andrew Wakefield figured out how. The tobacco companies figured out how. If you mean more explicit mind control, of course it's possible. It's simply that we don't know enough about how the brain works to be able to design techniques. I guarantee you in, oh, 500 years (or so) we'll know enough to make mind control chips technically feasible. $\endgroup$ – Ton Day Dec 31 '19 at 12:40
7
$\begingroup$

No

The problem with creating a "unhackable" system is that it would solve the Halting Problem. The problem asks if it is possible to build a system which takes some source code as input and outputs whether that code will stop eventually or if it will run forever. It was proven that such a system would be impossible to build.

We can reconfigure the halting problem to search the source code to see if it will ever execute illegal instructions such as unauthorized data access or modification and you can see that perfect virus scans are essentially the same thing as a halting problem system. That is to say, impossible to build.

Never have we mentioned what material the system is built from, electronics, software, or neurons. That is because it does not matter what material you use, the halting problem was proven impossible in the realm of logic and therefore proves it cannot be done in reality.

The only thing biological components will do to your system is:

  1. Make it more expensive to build your system, since you will need engineers experienced in traditional electronic software as well as "wetware"
  2. Increase the maintenance cost for the previous reason, plus because wetware degrades, gets sick, becomes tired, etc
  3. Decrease the performance of your system, because wetware doesn't communicate quickly and because the two systems will have to translate all data between the different formats used
  4. Increase the number of production issues caused since, presumably, wetware is less predictable than hardware and thus programs run on my system may behave slightly different than the same program run on yours
  5. Increase the political controversy around your product as people ask if the system is alive and if it is cruel to utilize it.

Overall, wetware is a more expensive, difficult to maintain, less performant and controversial way to achieve not as much as what traditional systems can do. The only advantage they might have is parallel processing, but even that is being done fairly well with traditional systems.

So, no.

|improve this answer|||||
$\endgroup$
  • 1
    $\begingroup$ "the two systems will have to translate all data between the different formats used"... This, to be fair, is also a potential hardening mechanism, but it can be implemented with or without biological components. At best, depending on how you employ it, wetware might force you to put more work into this sort of hardening... but that's almost like saying an acoustic modem is more secure than optical fiber because it inherently rate-limits brute forcing attempts; not exactly false, but still stupid. $\endgroup$ – Matthew Dec 30 '19 at 21:10
  • 1
    $\begingroup$ "reconfigure the halting problem," c.f. Rice's theorem. $\endgroup$ – 2012rcampion Dec 31 '19 at 7:54
  • $\begingroup$ All the downsides you mentioned could be solved with more reserch in that field and are only temporary. When we switched from mechanical to electronnical computing, electronical computing had just as much downsides at first, but trough reserch it was able to defeat mechanical computing. Your comment is todays equivalet of "Transistors are way to difficult to manufacture and expensive, lets stick with vaccum tubes!". I agree with you about the hackability tho. $\endgroup$ – SinOfficial Dec 31 '19 at 13:23
  • $\begingroup$ @SinOfficial We've been researching and trying to get wetware systems to perform better. Before computers were invented, all offices were wetware systems with paperwork storage. Despite mental athletes trying to calculate as quickly as humanly possible, we still can't keep up with electronics. $\endgroup$ – Muuski Dec 31 '19 at 17:51
  • 2
    $\begingroup$ I do not belive that this is true. Of course, you can't prove whether an arbitrary code would ever execute a forbidden instruction, much like you can't solve halting problem for an arbitrary Turing machine. But you can solve the halting problem for some specific turing machines and their specific inputs. Same way you may be able to solve "forbidden instruction" problems for some specific programs. You can definitely make a system that allows only programs it can prove to be safe. How powerfull or hard to program for would such machine be is an entirely different problem. $\endgroup$ – Viki Dec 31 '19 at 21:51
9
$\begingroup$

It is not true that human brain "has never been hacked". Humans have been made to behave against their true intentions, do the things that actually would never be doing on their own - just by being cheated, seduced, misinformed, provoked and the like.

Well known examples outside the scope of the "human brain" are impressive: a virus turns all the complex machinery of the infected cell just to produce more viruses.

Hence the initial assumption is probably wrong: it is not unusual for one biological system to "hack" another, it happens all time and not just between humans. From the other side, there is no reason why neural networks could not provide additional protection against hacking, same as any other advanced technology.

|improve this answer|||||
$\endgroup$
  • 1
    $\begingroup$ A simple example of such hacking would be some classes of optical illusion, such as Kanizsa's Triangle which shows something that isn't actually there. $\endgroup$ – Mark Dec 31 '19 at 22:31
  • $\begingroup$ Since the OP is asking about the transmission of data along biologically based pathways (rather than data interpretation or decision-making), it might make sense to also mention things like the way eyewitness statements are often inaccurate since what people think must have happened shapes their memories of an event, or the way taste or smell is very easily influenced with people reporting scents or tastes not present because someone said it should be there, or other variations of the placebo effect - these all show the unreliability of data transmission or storage within biological systems $\endgroup$ – Megha Jan 8 at 8:16
1
$\begingroup$

No.

The best protection is air-gapping; don't allow untrusted information into the system in the first place. The next best protection is to thoroughly vet incoming information. This may take the form of requiring incoming information to conform to a very specific format that minimizes the potential attack space. (It's a pity Roland Hughes, AFAIK, isn't part of the community; he has a ton of interesting stories in this area.)

|improve this answer|||||
$\endgroup$
1
$\begingroup$

We already have this!

A primitive human could benefit from a canine's keen senses and biological weaponry. The dog can anticipate some threats, deal with some itself, and in other circumstances buy the human some time for a different and more robust response. That is why our species domesticated dogs. Or the dogs would probably turn that on its head and point out how humans can anticipate and solve problems beyond the abilities of dog, and so humans are valuable additions to the pack. A smart dog will take its cues from the human and at the end everybody eats.

Computers use human brains right now. I install protective software on the computer. I decline to open suspicious emails and websites. I am careful what I plug into my computer in case it has malware or a virus. I can perceive and act on threats to the computer. And all the wetware I use on the computer's behalf comes with build in apparatus to sustain it, like my circulatory system and gut.

A computer can definitely benefit from a human associate. Humans are unsurpassed at perceiving order in chaos, rapid pattern recognition, suspicion and strategy. The way to use biological neurons to help defend computers is to have human beings doing the job, like we do.

|improve this answer|||||
$\endgroup$
0
$\begingroup$

Would a computer using biological components as an interface be any good?

Also can using biological components provide some other advantage to a computer, like flexibility or better pattern recognition?

The Terminator movie series gives a tangible example of where biological components (i.e., living skin) could be utilized, albeit not for the betterment of mankind. It's certainly conceivable.

Biological entities are generally able to reproduce, so biological components could plausibly self-improve over time.

In computer gaming, it's often more interesting to play against humans, so a biological AI might be interesting for gaming (and may even have military applications).

As another point along these lines: 3D-printed organs.

Would that protect it against hacking?

So would using neurons to transmit and translate information between the processor and the transmitter/receiver get rid of the risk of a virus getting through?

It would introduce a lot of incompatibilities, so things like today's computer viruses and trojans would not be useful. But if it's possible to create a computer-biological interface, and a hacker has access to it, then it's plausible to trick the computer in some way.

If it were like a human brain, then they're already able to be manipulated in various ways.

Or provide more complex processing?

This is plausible. Brains have abilities that computers generally don't have, such as the ability to learn from experience. But they also make mistakes.

If it doesn't then what would be a good alternative?

The only real way to prevent hacking is to isolate the computer, such as with missile launch systems. And even then, it might still be physically hackable if someone were to do so in person.

Or would it be too much of a nuisance to be any help?

It's hard to say. E.g., if a brain could be hotwired to perform computations, it might massively outperform a pure computer and have benefits such as learning and growth (but it's a big "if").

|improve this answer|||||
$\endgroup$
0
$\begingroup$

Also can using biological components provide some other advantage to a computer, like flexibility or better pattern recognition? Or would it be too much of a nuisance to be any help?

Human brains are biological computers that are superior to silicon ones at many functions (yet infirior at many others). One could argue that it would make sense to develop a biological computer to make use of those adventages.

I would challenge that. I'm no expert at this but afaik the main differences between computer and human brain architecture are that:

  1. brains are analog while CPUs are usually digital.
  2. information processing is much more parallelized in case of brains

I see no reason why those qualities couldn't potentially be achieved with silicon-based architecture but there might be one.

Having said that, I see some really good potential adventages of biological CPUs:

  • ability to self-repair
  • ability to change it's own architecture at will
  • ability to self-replicate (would make production very cheap, could also allow for autoscaling)
|improve this answer|||||
$\endgroup$
0
$\begingroup$

Memory would actually be vastly improved by biological interfacing. According to a May 2010 Scientific America article, the human brain is capable of memory storage of 2.5 petrabytes of information (or 2.5 million gigabits... most commercially available computers have a single or double digit gigabit storage). To give you an idea of how much digital information this is, if you you were to have a device with the memory storage of a human brain attached to a TIVO and set to record your favorite channel until memory was full, it would record 300 years of broadcasts before it was filled up.

It is important to know that this still doesn't equate to actual biological memory (i.e. the stuff you can recall) as there is no way to convert your fond childhood memory into a quantifiable digital memory value (a byte). But the storage capacity of a human is still thought to be unable to be met in a human life time.

|improve this answer|||||
$\endgroup$
  • 1
    $\begingroup$ "most commercially available computers have a single or double digit gigabit storage" — Uh... did you mean terabytes? Gigabytes would be correct only for RAM, and I suspect that 2.5PB figure is more closely related to attached storage than "working memory". $\endgroup$ – Matthew Dec 31 '19 at 16:31
  • 1
    $\begingroup$ My computer from 15 years ago came with a triple digit gigabit storage - 320 to be exact (40 GB). Nowadays, people are getting triple digit gigabyte SSDs for just OS + applications and then few terabytes of storage as a secondary (or external) drive. A petabyte of storage is not even too far fetched for personal use - some people have that much for their computers. It's still out of the ordinary for most but it's not really that shocking if somebody says "Oh, yeah I do a lot of video editing, so I have a petabyte of storage" or something. $\endgroup$ – VLAZ Dec 31 '19 at 20:50
  • $\begingroup$ @Matthew: Thanks... meant RAM. $\endgroup$ – hszmv Jan 2 at 13:39
0
$\begingroup$

No.

The network is there to send and receive information. This information has to be translated through programs to make them useful for you. So if you download a game first a program (like the newest Wyndowz) has to translate it and make it into its own program to run.

If a biological system first processes it... what remains of the program? Unless the biological program only processes visual information and presents it in the form of a video or something its not going to be any use. Even then if you convince the biological programs to add in a feedback loop where it shows just dicks over and over again and is too busy to accept new information its still a virus. A virus is basically nothing more than a program that does something you don't want. So if the brain prevents a virus from running then it would also prevent you from downloading any other program.

|improve this answer|||||
$\endgroup$
0
$\begingroup$

Would using neurons to transmit and translate information between the processor and the transmitter/receiver get rid of the risk of a virus getting through?

Yes. What you're describing is a technique already implemented without using biological neurons attached networks called "sandboxing". You can create isolated contexts that are only allowed access to the resources set-aside for them. Cellphones use them for apps. Even web browsers use these with CORS (cross-origin resource) limitations. Any application can run in the system, but it is required to live within those limitations.

Many previous answer cited the Halting Problem as proof that it is impossible to absolutely determine the behavior of a system allowing any input. It is true that for a program that allows ANY input, and when you want a absolute solution. But this is, in my opinion, a mistaken application of the Halting Problem. You could cite Zeno's Paradox as proof that you can not cross the room, but would be a similar, in my opinion, mis-application.

If you are willing to disallow certain inputs (perfectly acceptable in a security context), and you are willing to settle for "good enough", you can get pretty good answers.

You can also ring out complex attacks by translating across machine languages : in-and-out of analog signals, in-and-out of some human-readable representation using a limited character set (XML without CDATA).

There is also a security technique called "air gap" security, which is a low tech version of exactly what you propose. In an "air gap" there is a human operator servicing requests sent in on the network. If everything looks right to the human operator about the request, he or she will go to a completely disconnected other system (paper filing cabinet, computer in a different room, or different computer at same desk with all its own peripherals and unconnected to the outside world) to get the answer. Answer in hand, the operator will then type in the answer to question without the outside-connected computer ever coming into contact with the secure systems (not even a USB drive)

As a historical side-note: air gap security is what Mrs. Hillary Clinton defeated when she ordered staffers to hand-copy top-secret data from an air-gap protected secure system to her personal one.

Can biological computers AKA brains not be hacked like computers?

Unfortunately, brains can be hacked in a different, but similar way to computers. "Brain Games" is a good series devoted to showing examples, but there are several other excellent shows.

Social Engineering (hacking the human beings) is, I've been told, part of most successful system breaches.

However, human brains are excellent at pattern recognition and classification. This is why neural networks (which are software designed to mimic the human brain) is very good in this role. Pattern recognition is used in anti-virus software to identify "suspicious" software, and is used on higher-end network routers to identify and handle "suspicious" activity.

It should be noted that a lot of what is being sold as "AI" on the marketplace these days is basic line-fitting services like Microsoft Excel had in the 1980s. Linear regression is different than classification, and classification is different than general intelligence.

General intelligence is a set of software designed to mimic the human brain that also has a few days, weeks, months, years of actual experience being "taught" by a "teacher" (which may be a fast-moving program with questions and answers). A general intelligence, like a human monitor in the same role, is looking at a much bigger picture for anything unusual. General intelligence may include procedures so that it can formulate explanations (hypothesis) for unusual behavior, and test those ideas by making changes, or dig in and do it's own research; consulting a broader database of possible explanations. A general intelligence may even manage it's own "focus" and "attention", performing time management to pay attention to the parts of the problem that appear most likely to yield results. It can also take unilateral action in seemingly dire circumstances (closing a connection or shutting down and isolating a system), or ask for help in urgent situations where it feels it's not going to be able to figure out what's happening in an acceptable amount of time.

|improve this answer|||||
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy