Now this is not to say that the EX virus COULD NOT do something like this being hyperevolved and coming from hyperevolved beings but as a general consensus about transhumanity. So far it is beyond our technological reach, all things considered.

I suspect that such a "cloud brain" would be terribly slow. It might possess vast power, but delays on brain-to-brain paths would be massive, as well as limitations of computational speed on the brains themselves (Heat limit - you can't go above 42 Celsius or proteins will denaturate; ATP transportation efficiency is limited; you will run out of chemical neurotransmitters if you use them too rapidly; myelin sheaths will dissolve because of the neurons firing too rapidly, etc.). The best decision made by newly awakened biological Seed AI might be to switch from wet-ware to hardware as soon as possible. But it's not necessarily a bad thing story wise. This state of powerful but slow intelligence my be desirable for the beings who use it: - they want to slow themselves down and wait for a "better tomorrow". - they want to solve some kind of problem that needs certain data collected over long period of time - for example orbit eccentricity of some group of celestial bodies; - they prepare themselves for non-FLT space travel to another star system; - this is their way to hide from other tranhumans - they might live in Jovian Republic and be afraid of being erased; - they just enjoy this kind of "spirit communion"; - they want to switch to non-organic gear but don't have the resources right now. OR they got infected by Watts-MacLeod and they are using some sort of psi effect to speed themselves up - but at the same time they are a kind of psionic array, very susceptible to outside influences. (And I mean OUT-outside - like 'some ETI or TITAN bastard-child living in the starless void' outside).

Biology is very slow compared to electronics. Signals move at 1-100 m/s, neurons fire at most a few hunded Hz and are at least several microns wide. Optical computers can move signals at near lightspeed, nanoelectronics switch on femtosecond timescales and can have nanometer-sized processors. While biology can be improved, there are fundamental limits to what neurons can do: ion channels, membrane potentials and similar stuff does not scale well. That said, I think one can do biological superintelligences. It is just that it is a fairly inefficient way of doing it with no clear advantage except backwards compatibility and maybe aesthetics. Here is what I can imagine some biotech singularity seekers to do: culture a *lot* of boosted cerebral cortex in tanks, linked to morphs and normal brains using mesh inserts. Use localized training to format it into useful cortical maps and behavioural loops. Enhance it via neurachem, multiple personalities and multitasking. It might not be as sharp as a software seed AGI, but it might have nice associative abilities. Of course, adding AGI coprocessing might make i even better once it has grown into a smart mind. Maybe the goal is to first make a solid biological superintelligence as template for software intelligence.

There are plenty of examples (in literature) where a biologically-based computer is considered, if not more efficient to an electronic one, a great booster component. Also, it is usually regarded as being able to make lateral thinking or non-linear deductions (jumping from fact A to conclussion D without solving conclussions B and C). Also, there is a difference between electronic AIs and biologically-based intelligences, and that would be the chemistry involved: an organic brain works both with electricity and chemicals, while a purely digital one only uses electrical signals. This being said, we need to consider the thread's target: "biological SEED AI". So, is it possible to develop an artificial intelligence able to self-improvement the same way an electronic one can? As usual, we need to state a difference between Ego and Morph, or software and hardware. This being said, it stands to reason that a biological seed AI would require much more time to increase or improve its hardware, regardless of the improvements it can make to its software: essentially, the need to grow biological components is a bottleneck. This suggests to me two scenarios: First, a group of transhumans in Eclipse Phase might find the idea of a seed AI whose improvement speed is somehow limited attractive, thus deciding to use it as a testbed for Seed AI research. Watts-McLeod capacity might come as a bonus... Which leads us to the second scenario, where the Seed AI somehow managed to integrate the research group into a mass of biomechanical connections, using them to boost its performance (essentially, adding biomorphs as "expansion modules" for both memory and processing power). This rather disturbing behaviour might point to exurgent activity, thus granting Firewall's attention. It comes to mind the possibility of a trap, where a bogus biologival Seed AI and some empty morphs are staged in order to attract a group of agents who, in case of knowing they will face a biologically-based async-exurgent threat, might decide the use of synthmorphs to be a wise option, allowing the trapperts to use some synthetic neutralyzing tactics to capture the team.