Fact-checking Ray Kurzweil’s 1999 predictions: second set
Part of an ongoing back-and-forth with KurzweilAI’s editor about the predictions in The Age of Spiritual Machines (1999), reviewed against the state of the world a decade later. Numbering picks up from an earlier round; predictions 18 through 26 cover face recognition, 3D chips, distributed computing, multi-core CPUs, nano-scale machines, and brain reverse-engineering.
Thoughts on the second set: 18.
While face-recognition technology is both available and widespread, I’d hesitate a little before saying that “computers” are reliably able to identify their owners, since most computers either lack a camera, lack sufficient resolution, or lack the software to do so. Still, Toshiba offers 119 different models with login through face recognition as an option as of May 2nd, 2010 ( Toshiba face-recognition guide ), so it’s essentially correct.
19.
We discussed this one earlier, I think. The original text of the prediction was: “In terms of circuitry, three-dimensional chips are commonly used, and there is a transition taking place from the older, single-layer chips.” By “older, single-layer chips,” Ray is clearly referring to 1999-era chips — which also had multiple layers of silicon and metals stacked on top of each other (see, e.g., this archived March 2000 description of how silicon chips are made ). Hence, the predicted feature in question can’t be multiple layers of silicon (since these were present in 1999); the most obvious interpretation is “multiple computing elements integrated vertically,” which, as we discussed, isn’t scheduled to hit mass production for most applications for another 3–5 years.
20.
I’d also mention the very common use of earbuds, which are very small devices that can duplicate most of the features of old large stereo speakers.
21.
I’d also mention the PlayStation 3, which is probably in much wider use than these graphics cards ( 33.5 million sold as of Sep. 2009 ), and has ~200 GFLOPS of CPU power and ~1.8 TFLOPS of GPU power .
22.
I’d also mention how the estimate of “human brain = 20 PFLOPS” was obtained, since a lot of people would question that.
23.
Distributed computing has been around since 1996, but the networks today are roughly 4 OOM more powerful than in 1999 (~5 PFLOPS as opposed to ~0.5 TFLOPS; see Entropia primenet status, archived 1999 ). As to Google, Yahoo, Microsoft, etc., I’m skeptical of Ray’s earlier claim that the Google server cluster can be counted as a single supercomputer (“ Kurzweil says he considers all of Google to be a giant supercomputer and that it is, indeed, capable of performing 20 petaflops ”), since these computers do a very wide variety of different tasks (as opposed to one single task, like searching for alien intelligence or folding proteins), and they are not controlled from one single location (as opposed to projects like BOINC, which provide central servers from which all work is distributed).
24.
One of the really big things here is that most CPUs nowadays feature multiple parallel cores, as opposed to the single cores present in 1999. Since most of these still only have two or four cores per chip, it’s completely correct that “most computer computations are still done using conventional sequential processing, albeit with some limited parallel processing.”
25.
I second Robert Freitas, and would add that the nano-scale features in microprocessors or nanotubes are clearly not what the prediction intended to describe, since it talked specifically about “autonomous machines.” Given the trends in technology today, I think the first we’ll see of nano-scale autonomous machines is custom-engineered proteins. We already have the technology to do DIY custom-built proteins; it’s just a question of cracking the protein-folding problem so we can construct proteins that do new interesting things.
26.
Both MRI and reverse-engineering the brain long predate 1999, so this prediction is sort of “true by default.” fMRIs to look at human neural activity have been in use since the early 1990s. The destructive brain scans on the deceased inmate that the text mentions were done in 1994 ( Visible Human Project ), and work on reverse-engineering the human brain has also been going on for a while (see this Discover piece from December 2002 ).