A superintelligence is a theoretical smarter-than-human being. Despite the superficial implausibility of the idea, numerous scientists have not ruled out the possibility of a superintelligence being created within the next century or even the next 20 years. Some researchers have even stated it as their career goal. Some of the most interesting speculations regarding superintelligence involve scenarios where the superintelligence continues to further enhance its own intelligence and capability. Such an event has been termed a "Singularity" by theorists.

Recent movies such as The Matrix have popularized the notion of a superintelligence with the public. Artificial intelligence (AI) is one of the possible paths to superintelligence, and perhaps the most frequently discussed. With robotics as its "hands" and computers as its "brains," a sophisticated artificial intelligence could think more thoughts and achieve more actions than a human would be capable of. This would prove especially true if the AI could redesign its own cognitive architecture, engineer and fabricate new robotics tools, and absorb more computing power. Many philosophers have expressed concern that such an entity could lack moral common sense and be a threat to humanity.

Other possible paths to superintelligence would be through human genetic engineering or Brain-Computer Interfacing. Brain-Computer Interfacing has improved to the point where mouse cursors and prosthetic arms can be controlled by thoughts alone. The military is pouring millions of dollars into developing the technology. Eventually we might create human beings capable of interfacing very closely with the global Internet, to the point where it would be difficult to distinguish between the two. This possibility has been explored in the movie and series Ghost in the Shell.

Human genetic engineering is probably the least likely path to lead to superintelligence, as such research on humans is heavily restricted in most developed countries. This path would take the longest to produce results - while a modification to an AI program or brain-computer interface might take only a few hours or days to test, a genetic modification would take 10 years or longer to test - the time it takes for the genetically engineered embryo to grow into an infant and then a child and adult. For this reason it is most likely that superintelligence will eventually emerge from Artificial Intelligence or Brain-Computer Interfacing research, possibly only within a few decades.

One additional possible route to superintelligence is the use of pharmacological or genetic therapies to boost human brain functioning. These approaches are less "radical" than Brain-Computer Interfacing, and therefore may have less power to truly augment intelligence, but their lesser degree of invasiveness may make the approach more acceptable for research and experimentation.