The Human Mind Known Knowns: Understanding Smart Technology Part 3

Published on INFORMS Analytics Magazine (Joseph Byrum)

Author’s Note: This blog series Understanding Smart Technology – And Ourselves examines our relationship with advancing technologies and the fundamental choices we face. As we stand at the threshold of an uncertain future shaped by artificial intelligence, the author challenges readers to consider whether we should embrace these transformative changes or resist them in defense of our humanity. Drawing from historical patterns of technological adoption and resistance, the series promises to deliver nuanced perspectives on our technological trajectory, beginning with a comprehensive overview of our current understanding of smart technology and its implications for society. Read Part 2 where the author discusses the “known knowns” of the hardware and software aspects of smart technology.

The human brain is the next stop on our ongoing exploration of the “known knowns” about smart technology. 

Properties of the Human Brain (Neural Networks)

In its physical properties, our brains serve as the “hardware” for our thoughts. At the same time, our brains share properties of “software” in the sense that the hardware itself generates pathways for intelligence. Harvard’s Connectome Project estimates that our brains have “tens of billions of neurons connected through perhaps one hundred trillion synapses” [1]. These pathways, known as neural networks, have been the subject of scientific research in both brain science and computer science for seven decades.

Most recently, a team of scientists have learned from functional magnetic resonance imaging data (direct observation of brains) that regions of the brain “tend to coordinate by forming a highly hierarchical chain-like structure of homogeneously clustered anatomical areas” [2]. The researchers also found “the hierarchy of network segregation and the level of clusters integration as a function of the connectivity strength between brain regions” [3].

The greater our understanding of the brain’s neural networks, the better we can be at fashioning computer networks that operate in the same way. But brain science is still relatively young, and there is still much to be discovered. 

Properties of the Machine Brain

Unlike with the human brain, the basic properties and information architecture are well-understood for computers. Modern digital computers are built on millions of simple electric circuits that can be switched either on or off (the binary system of zeroes and ones mentioned in Part 2 of this blog series). At the circuit level, the functionality of computers is quite limited; computers can essentially only store, move and compare data.

All higher-level computer functions are built on these three functions, all of which use simple logic operations to accomplish their ends. Even simple addition is implemented in the binary system as a logic operation at the circuit level through logic gates. The logic gates do the comparing; they add numbers using a binary system and the logic of AND, OR, NOT and XOR (XOR means one or the other, but not both) [4].

What makes modern computers powerful is that they can perform billions of these operations in seconds. We use software programming to tell the hardware circuitry how to perform these operations. Most programmers work in high-level software languages that are designed to be understood by humans. Each line of high-level software code typically issues multiple instructions to the hardware after it is translated (compiled) into machine language.

Since software tells hardware what to do, hardware and software are logically equivalent. Doing a certain application in dedicated hardware is usually faster, but software with generic hardware is more flexible.

Neural networks are computer programs that mimic the information exchange architecture between neurons in the human brain. These multilayered networks are trainable by helping them understand what outputs are expected from certain sets of inputs. DeepMind co-founder Demis Hassabis argues that the only way for artificial intelligence to realize its true potential is to build it more and more on insights gained from deeper research into the functioning of the human brain [5]. 

Fundamentals of Human Behavior in Relation to Machines

We know that human beings make and use tools – physical objects intending to make our actions more powerful or efficient. Our oldest fossils show we were doing this 300,000 years ago [6]. Today’s machines are more advanced tools, and humans even form attachments to them – think of cars. We often see humanity in nonhuman things – everything that moves and breathes, from the coldest machines to the cutest creatures.

Machines can at times become a rival or even an enemy. Entire social movements have formed around protesting them. As a result, laws and policies have been created to govern the role of machines in society. 

Multiplier Effect of Technological Change

Meanwhile, all of these changes are happening so rapidly that their growth can be described as exponential. As noted in the website for Singularity University [7], a technology can be considered exponential if its power or speed double each year, or its cost must drop by half [8].

New technologies change how the factors of production may be combined to produce a particular level of output. The factors of production are usually defined as land, labor, capital and entrepreneurship. An improvement in production technology typically lowers the quantity of at least one of the factors needed to produce a given output, although it may increase one of the others. In smart technology, as in other economic phenomena, there is “no such thing as a free lunch.” Positive change in one domain can bring dislocations and distress in others.

Today, all of these forces are operating at an astonishing speed, and it takes time for humans to adapt to big changes. We need time to adapt. This involves the passing of the old generation set in the old ways, to a new generation comfortable and proficient with the new ways. It may well be that the biggest challenge posed by the Fourth Industrial Revolution is not the magnitude of the changes it is bringing about, but the rate at which these changes are happening. Economists are normally sanguine about technological disruption because textbooks say the old jobs that disappear will be replaced by yet-to-be-imagined better new jobs. It’s not surprising to see some are concerned that the change is now happening too fast.

Bringing it All Together

We know that smart technology is based in computer science, and in particular the algorithms of computer programming. We have flexible and useful notions of human intelligence, and we understand the chemical components of organic life – something no smart machine can ever possess. The scientific community can tell us about the nature of the human brain and the machine brain. We also know a fair amount about human behavior in relation to machines. In addition, we have a good understanding of technology economics and its multiplier effect, posing challenges to humanity’s adaptive skills.

References

1. http://cbs.fas.harvard.edu/science/connectome-project
2. Rossana Mastrandrea, et al., 2017, “Organization and Hierarchy of the Human Functional Brain Network Lead to a Chain-Like Core,” Scientific Reports, Vol. 7, Article No. 4888, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5501790/.
3. Ibid.
4. http://improve.dk/adding-67-at-the-logic-gate-level/
5. Demis Hassibis, et al, 2017, “Neuroscience-Inspired Artificial Intelligence,” Neuron, Vol. 95, No. 2, pp. 245-258, http://dx.doi.org/10.1016/j.neuron.2017.06.011.
6. Ann Gibbons, 2017, “World’s oldest Homo sapiens fossils found in Morocco,” Science, June 7, http://www.sciencemag.org/news/2017/06/world-s-oldest-homo-sapiens-fossils-found-morocco.
7. “An Exponential Primer,” https://su.org/concepts/.
8. Ibid.