by John C. Lennox 2020
Book Review by Ray Herrmann
The author's premise: AI (Artificial Intelligence) is a set of specialized tools that can process large amounts of data and extract solutions much faster and better than humans. AGI (Artificial General Intelligence) is man's attempt to create a superior life form (super intelligent robot). But Lennox says AGI will not be conscious. Furthermore, its designers (programmed by individuals) would have power over us as they are the ones that set the rules. And because they set the rules, this story is more like a half-factual, documentary of the plausible near-future. They, and not AGI, are really in control.
The current uses of AI to persecute and control people (mainly in China) are mentioned as some threats AI poses (like many tools, they could be used to help or to harm). But, although AGI seems to pose the greater threat, its successful development is seen as unlikely for reasons that seem to often touch on God (but I see these same inherent restrictions without the need for a God).
The author is well read and he cites many, many references, both pro and con, but as a devout Christian with prolific studies and writings, he gets drawn into parochial interpretations of AGI. I feel compelled to respond to many of his arguments, but, for brevity, I will post only a select few. Below, find his statements or apparent meanings (in back text), followed by my opinions (in purple text).
pg 115: The book of Genesis begins with the biblical narrative that makes most sense to me, of God, the universe, and human life. "In the beginning, God created the heavens and the earth." This tells us that God is primary, and the universe derivative. Matter is not the only reality; it is not even the prime reality.
Rather than start by assuming a timeless all-wise God, I suspect that the universe - any universe - has to start with nothing. Otherwise there is the circular question of "So, where did that come from?" There is even some evidence that hints it may be the case that Space, Time, Matter and Energy are constructs, "real" only within themselves:
That Space and Time are internal artifacts of our universe is hinted to by the finding that, at the scale of atoms, we notice "Quantum" affects, meaning that we find two particles somehow linked to each other, even though they may be miles apart (know one and you know the other!). Distance temporarily looses meaning! And "Time" is just an artifact of change-over-distance. But at our very large scale of existence, reality is statistically smoothed such that things seem stable to us and seem to conform to behavior rules (laws of motion, position, mass, charge).
pg 123: We are told that the source of life is the breath of God. Getting some kind of understanding of the relationship of the conscious mind to the body is acknowledged by all to be very difficult. Matthew 10:28 shows that the soul is now understood in this text, to be the living body, as it cannot be killed by humans. …there is a non-physical aspect to human beings.
I suspect the formation of our "consciousness" is an outcome of natural evolutionary pressure, and to some extent, not even unique to our species. Scientists have used the "Mirror Test" to signify self-awareness (Conscious ones see themselves, whereas not-conscious ones see another critter). Studies also found some self-awareness (consciousness) in Dolphins, Chimps, some Birds and other species.
pg 124: It is claimed by the author that AGI cannot be created because we humans are unable to program "curiosity". Computers don't have goals of their own.
I tend to agree that we can never create a human-like "Artificial General Intelligence", because to do so, we would have to endow this machine with the same needs that drive humans. Things like, the motivations of pain and pleasure, tastes to distinguish poisons from nourishments, hunger to assure gathering energy for growth and on-and on. Surely we could design a machine that didn't need or seek some of these things, but then would it act "human"? And that AGI machine we built would be composed of only those motivations that the human designer programmed in! Not transcendent at all!
I think Nature developed mental attractions and avoidances ("curiosity") through trial and error. The species that coincidentally developed more survivable attributes faired better than those that did not and thus, over time, they replaced those. The Earth was formed about 4.5 billion years ago. But "life" seems to have formed a mere 580 million years ago (implying a great difficulty, even given the vast regions of favorable conditions on earth). Yet since that time, earth has exploded in variation, implying the power of self-replication, likely occurring with times of acceleration and times of decline as competing forces were accommodated. All the while, Natural Selection pitted current species against others present and against the local environments.
pg 152: The fairly obvious fact that there are many influences involved in human choice does not rule out the fact that there is sufficient freedom for morality to make sense where, in the absence of that freedom, it wouldn't.
First, this is an excellent example of how the author typically stages his arguments by claiming certain conjectures as "facts", and then proceeding to build his case on them. To me, this is like building skyscrapers on sand.
Second, just because there is enough uncertainty in our understanding the elements of free will ("human choice"), does not justify one injecting one's favorite attribute (morality here) into the mix.
Personally, I think free will is an illusion. Evolution has endowed our brain with the ability to consider many relevant inputs at any moment and, by assigning importance to each factor, bring these quickly to summation to direct our next move (acting fast and correctly is very survivable). Some common factors we integrate are: sensory inputs (pain/pleasure, hunger) desire, logic, projected outcomes, goals and our current state of mind. We probably develop each decision by weakening or strengthening the importance of each attribute in accordance with our ability to reach intended goals. As we learn to deal with conditions over time, we form our personality (which seems to solidify with age). This is where parents (who have more mature outlooks) help in the development of their kids (who are just forming).
The Marshmallow Test has been used on 5 year-olds to predict their later success in life. Addiction to drugs is an example of our "free will" being bent by adding weights to some urges. How hard is it to quit smoking? …to stop eating? Notice that as one "matures" with age, one's brain learns to give more importance to long range predictions whereas babies respond mainly to their sensory inputs.
What goals would an AGI machine use to make decisions? I suspect they would all be just whatever was programmed into them. But would or could they develop self-motivations?
For those who believe in God and want some really great arguments for it, I recommend reading chapter 11 (pages 173-194). The author even makes plausible arguments (assuming you accept his "facts") for the return of Christ and even for the ascension of the remaining people.
This whole book is peppered with arguments against the writings of Yuval Harari (author of Sapiens, Homo Deus …). Of course, the universe is large enough to contain many theories about its construction and its motivations. That leaves us free to find support for many beliefs as we prefer.