Khan Academy announced this morning that it has raised $5 million from the O'Sullivan Foundation (a foundation created by Irish engineer and investor Sean O'Sullivan). The money is earmarked for several initiatives: expanding the Khan Academy faculty, creating a content management system so that others can use the program's learning analytics system, and building an actual brick-and-mortar school, beginning with a summer camp program.

The $5 million marks the latest in funding for the non-profit, which has received over $2 million in grants from the Gates Foundation and from Google.

Part of the lure of the Sal Khan narrative is this idea that he is single-handedly educating the world through his self-made YouTube videos. No doubt, you can point to page and video views to make a case about his impact.

Khan has long kept full control over the "instruction", or rather the video creation -- all the content has been created by him. That changed last month, as I reported here, when Khan Academy struck a partnership with SmartHistory, bringing on that organization's Dr. Beth Harris and Dr. Steven Zucker as art history instructors. The money from the O'Sullivan Foundation will be used in part to expand the Khan Academy further, to at least 5 full-time-equivalent teachers.

The O'Sullivan Foundation grant will also be used to build what's described as "a crowd-sourced content management and curation system." Details are sketchy on exactly what this entails, but the press release compares it to Wikipedia, saying it's a "similar outlet for dedicated professionals to develop quality instructional content." The system will also enable others to tap into some of the tools and analytics that Khan Academy is developing.

Khan Academy intern David Hu offered some great insight this week into what these analytics look like. In a blog post entitled, "How Khan Academy Is Using Machine Learning to Assess Student Mastery," Hu detailed the efforts underway at Khan Academy to rethink how its model for student proficiency works. Currently, it relies on a "streak" -- that is, students must get a certain number of questions right in a row in order to move on. Hu proposes an alternate approach to ascertaining whether or not a student has gained proficiency (defined as a 94% or greater likelihood of correctly answering the next question asked involving that skill) using a logical regression model. Hu hypothesizes that with this new proficiency model, learning outcomes should increase, in part by moving students off of problems that they're good at more quickly.

With its current level of funding, no doubt Khan Academy has been able to attract some real super-star engineering talent to its team -- a team that has remained fairly small. There are, I think, under 20 employees, including the recent SmartHistory additions. But it's worth noting that while the engineering brainpower is sizable here, the number of teachers (past or present) on board is small.

"Teachers don't scale," I remember Sal Khan saying to me when I interviewed him last year. What can scale, he argues, is the infrastructure for content delivery. And that means you just need a handful of good lecturers' record their lessons; the Internet will take care of the rest.

But online instruction clearly isn't enough, and as "blended learning" becomes the latest buzzword -- that is, a blend of offline and computer-mediated/online instruction -- Khan Academy is now eyeing building its own school. The money from the O'Sullivan Foundation will go towards developing a "testbed for physical programs and K-12 curricula," including an actual physical Khan Academy school. This will begin in June 2012 as a series of summer camps.

"The school of the future will not resemble the school of today," Khan says. "In the past, the assembly-line, lecture-homework-exam model existed because that's what was possible in the no-tech and low-tech classrooms of their day." His team now have $5 million to take that lecture-homework-exam model into the high-tech classroom� or something.