Singularity with a multi-electrode array.

[...]

This does not correspond to what Vernor Vinge denominated an "intelligence explosion" that the Singularity will involve. While pretty awesome, the Frankenrobot has "50,000 to 100,000 active neurons", whereas a human brain contains roughly 100 billion neurons, many of which, such as the 80,000 spindle cells which control high-level emotional responses, are highly specialized and not fully understood yet. In order to be able to achieve functional simulation of human intelligence, we need to fully understand the different parts of our mind and their networked interactions at least as well as we understand the cochlea. The step from rudimentary control of the movements of a robot to human levels of control (which involves 50 billion neurons for the sensory perception, motor control and coordination in the cerebellum) is significant.

Another problem with your brain-robot interface is that it doesn't bypass the biological limits of the human mind. For instance, computers can analyze hundreds of millions of positions on a chess board every second, whereas a human (such as Kasparov) can analyze about one position in the same time frame. The reason why humans are, despite our probabilistically inexorable odds, able to beat most computers at chess is due to our currently superior pattern recognition abilities (in most areas). In other words, we are infer knowledge about current events from past experiences. That being said, the chess rating of human grandmasters of chess is fairly constant, whereas the chess rating of computers was increasing exponentially for every year that passed, until Kasparov was defeated by IBM's Deep Blue in 1997.

While other developments have been made in algorithms for chess-playing computers after that point of time, it marked the end of specialized chess circuits (primarily CPU's), where were approximately a hundred times faster than its contemporary general-purpose circuits. However, Moore's law accurately predicted that general-purpose circuits would be as fast as chess-playing circuits before the end of 2004. Now, four years later, we're encroaching a point in which home computers will be able to beat human grandmasters in chess.

I figure it could be possible to utilize expert systems with built-in subsystems of both brute force and narrow AI approaches solutions for various problems. In other words, if the human version of "Frankenbot" that you're referring to would see a chess board via its pattern recognition algorithms, its systems could automatically start performing calculations of the different outcomes of prospective chess moves. The results could then be reported back to the human brain through comprehensible means (for our anemic, biological minds).

As for the intelligence explosion, even the initial inception of human levels of nonbiological intelligence will be insufficient to result in a technological singularity for some time. Although it will be able to think much faster than us and share information with other nonbiological, intelligent entities almost instantaneously, the first step toward superhuman intelligence is still expected to take, according to Kurzweil, as long as a decade due to the difficulties pertaining to its inception.