The Living Legend

One of nature’s more-robust solutions is in DNA, the chemical strands found in all living beings. DNA exists in several forms, most notably comprising long strings of nucleotides that appear in sequences unique to each individual. The DNA sequence of a creature— its genome—remains nearly constant throughout its life, and is evident in bones and other remains after death. Although the half-life of DNA can be as short as 500 years, a complete genome has been decoded from archeological samples as old as 700,000 years.

Some DNA defines a creature’s essential traits, such as eye color and propensity for certain diseases. However, there is also a considerable amount of nonfunctional “junk” DNA in which additional information could be encoded. Researchers have written in those spaces since 1988, when they successfully encoded a 35-bit graphic into the DNA of E. coli bacteria. In 2012, a team led by Harvard Medical School genetics professor George M. Church surpassed previous storage records by putting his 5.27-megabit book “Regenesis,” comprising 53,426 words, 11 images, and one JavaScript program, on a string of nucleic acids. Church believes his team’s accomplishment is only the beginning, saying this storage medium could theoretically store over 450 million terabytes per gram of DNA.

DNA storage has another advantage over traditional media: it could be made self-replicating within a living being,

along with Thinking Machines’ Danny Hillis—concerns itself with projects to “help make long-term thinking more common.” One of them is Long Server, an “overarching program for Long Now’s digital continuity software projects,” including an in-the-works file-for-mat conversion system. According to executive director Alexander Rose, “For us, the major element is an awareness one. No industry has really tried to make people aware that their data is ephemeral ... I would defy you to open up a current file 200 years from now in digital format.”

The Medium Is the Message

The Long Now Foundation’s Rosetta Project explicitly avoids the digital compatibility issue by micro-inscribing more than 13,000 pages of information, in more than 1,500 languages, on a palm-size disk (a zoomable online version may be seen at http://rosettaproject. org/disk/interactive). The data is like a printed page: analog, albeit legible only through a microscope. In this way it uses essentially the same technology as the original 2,200-year-old Rosetta Stone, as did the “Pioneer Plaques” launched skyward on spacecraft in 1972 and 1973.

The Rosetta Disk is made of solid
nickel; the Pioneer Plaques were made
of gold-anodized aluminum. As part of
his thesis, University of Twente doctor-
al student Jeroen de Vries explored oth-
er materials in an attempt to create an
even more durable data-carrying sub-
strate. He ultimately devised an opti-
cal disc of tungsten coated with a clear
layer of silicon nitride, whose heat
resistance he demonstrates on video
by cooking an egg on it. He estimates
the QR codes he inscribed on it would
still be readable after a million years at
a temperature of 200 degrees Celsius;
however, he acknowledged “the disc is
actually very fragile, because it is made
of crystalline silicon, so it breaks very
easily along the crystal lines.”
The approaches of de Vries, the Ro-
setta Project, and the Pioneer Plaques’
creators all are to freeze information
on a static medium. This approach is
comparatively limited: the oldest delib-
erate, data-bearing artifact is perhaps
a cave painting 40,000 years of age, a
mere toddler in the grand scheme of
things. If we are to present ourselves to
descendants beyond that, we will need
a better medium.

DNA storage has
another advantage
over traditional media,
it could be made
self-replicating
within a living being,
and be passed on to
offspring through
natural reproduction.

ACM
Member
News

BIAS, SUBJECTIVITY MAKE BOTH HUMANS, AI, INTELLIGENT: STEIN

“I think of
myself as an
applied
philosopher
and a computer
science
education
innovator,” says Lynn Andrea
Stein, professor of computer
and cognitive science and a
founding faculty member at
Franklin W. Olin College (Olin)
in Needham, MA.

Stein graduated cum laude in computer science from Harvard and Radcliffe colleges; she obtained her master’s degree and doctorate in computer science from Brown University. Artificial intelligence and robotics have been areas of specialization throughout her teaching career.

In 2000, Stein left her position as associate professor at Massachusetts Institute of Technology (MIT) to help found Olin, an innovative undergraduate engineering college that espouses hands-on engineering and design projects rather than a traditional core-curriculum teaching model.

“It was a once-in-a-lifetime opportunity to start a college from scratch and think about what education could and should look like,” Stein says.

“I researched how to understand
and create intelligent systems
and create systems in which
people can learn well.”
Stein believes bias and
subjectivity make both humans
and AI systems intelligent. “To
discriminate, in its original
sense of differentiation, is a
fundamental necessity for
human beings and AI,” she says.

“In that sense of discrimination,
we don’t want to rid ourselves
of the ability to differentiate.”
She applies these concepts
to her work in AI and robotics.

The robotic systems her research group builds bridge the gap between low-level behaviors associated with robotics and levels of cognition that better approximate human thinking.

—Laura DiDio