The Atlantic: super-brains and baby talk

The July/August 2013 issue of The Atlantic has two short articles relevant to our robot futures. From the Attention Dilution Disorder department comes “Papa, Don’t Text,” by Deborah Fallows. Fallows reports how babies’ speech development turns out to be sensitive to just how their parents interact with them. Several studies have shown that direct linguistic interaction between parents and babies is far superior to babies’ hearing adults talk to one-another, talk into the phone, or babies hearing the staccato clicking of a texting parent.  Another study Fallows summarizes even shows how a live person can talk in a foreign language to a small child and trigger language learning, whereas electronic versions of the teachers fail to cause the same learning. If my trips to the local park are any indication, smartphones certainly do a disservice to such authentic, two-way interactions between children and parents today; telepresence taken to the limit will, I believe, further reduce the amount of pure, two-way, high-quality interaction our future children will witness.

 

Second, in “The Intuition Machine” Alexis Madrigals conducts an interview with Jeff Hawkins, who made Palm a great success in the early PDA world and who is now developing super-intelligent neural network-based computer systems for mining big data. Jeff believes in designing future computer AI based on direct neural modeling: building the best possible simulation of the human neocortex that’s possible in a digital machine. Hawkins correctly identifies entire markets where these software AI’s will achieve superiority: exerting control over very complex machinery; forecasting power consumption with super-high fidelity; evaluating massive human behavioral correlations; etc. One telling example Hawkins gives is that such a system doesn’t just do ‘demographics’ to predict how 10,000 smart meters help predict tomorrow’s power demand. Instead, his system would model all 10,000 smart meters individually! This is the spirit of New Mediocracy: data understanding systems will model us in progressively smaller buckets and categories, until your future behavior is bought and sold for profit. Ironically, Madrigal brings up underemployment at the end of the article, asking Hawkins what people will do if machines become so intelligent. Hawkins responds by pointing out that the machines will do things humans never could, and then he jumps over to the tried and true AT&T switchboard example: “If we had to have an operator place every telephone call in the world, there would be a billion telephone operators.” The article does a good job of giving us readers some insight into the scientific motivation of folks who change the world, and just how they view the sociological impact they might have.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s