The spacesuited figure on the cover makes it look like science fiction, the opening liability disclaimer would fit right into to an end-user licence agreement, the index is sandwiched between a 40-page bibliography and a handy list of abbreviations, from CAPTCHA to YOLO, that you probably know but might like an reminder of, and the whole book is covered by a Creative Commons licence. So how does this unusual approach to publishing stand up?
The introduction to Exponential Progress purports to be written at the back end of the 21st century, reframing the usual ‘outhouse to zero-gravity toilet in 70 years’ reference to the speed of progress as a 300-year jump from no electricity to ‘outer-terrestrial colonies’ (along with hints about a civilisation of AIs).
By promising to explain to fictional readers some eight decades in the future how we got there, author Farabi Shayor gets licence to cover the state of the art across a range of current and bleeding-edge technologies — virtual reality, electric and self-driving cars, AI (both software and ‘brain-like’ chips), the singularity, brain-computer interfaces, CRISPR and synthetic biology — for a general audience.
Although the book promises to explore the dangers of emerging technology and whether the pace of innovation is beyond human control, the writing is often unstintingly optimistic. For example, Shayor presents cryptocurrency as ‘the money of the future’, where the only problem to be solved is an over-reliance on GPUs; he addresses neither environmental issues nor the inherent latency in blockchain ledgers. Blockchain itself is also depicted as the solution to safely sharing genetic information, while ignoring the questions of access by commercial and law enforcement organisations.
SEE: Building the bionic brain (free PDF) (TechRepublic)
Similarly, a fairly shallow account of the history of VR suggests on the one hand that it’s too early to build successful companies on, but on the other that all problems will be fixed in five to seven years. Meanwhile, Shayor’s potted history of EVs and self-driving cars spends far more time on the diesel emissions scandal than on self-driving accidents, or the fact that the delivery of fully autonomous cars keeps getting pushed back into the future.
Scientific advances like micro-algae fuel are routinely presented as if they were ready to adopt, with no examination of whether lab techniques will scale up for production: the assessment of why VR isn’t ready isn’t repeated for any of the other technologies covered — which will, apparently, just work.
The sections on CRISPR and synthetic biology include an interesting explanation of why lab-grown ‘meat’ tastes the way it does, a passionate attack on the costs of drugs developed by large pharmaceutical companies, round condemnation of how governments have handled the COVID-19 pandemic, and the potential of synthetic DNA storage (albeit with the odd suggestion that sensitive information could be stored by injecting it into people).
A high-level survey of machine learning or neuromorphic chips will necessarily be simplified, but still needs to be accurate. We like the analogy of standard light switches and dimmer controls to show the difference between digital and analog information, but some of the basic explanations of computing are just plain wrong, even allowing for the attempt to make technical concepts clear to a possibly non-technical audience.
Error correction needed
The author appears to fundamentally misunderstand the concept of the memristor, among other things, and asserting that deep neural networks (or possibly reinforcement learning systems) learn by understanding the difference between good and evil is at best absurdly anthropomorphic. The casual reader may also be left with the impression that all significant breakthroughs in almost every area of technology have come from Google, Tesla or IBM.
This book would have benefited from a copy editor to clean up some of the inventive but oddly-phrased idioms, and a technical editor to catch a range of errors — like suggesting that a 20MW Chinese supercomputer uses as much power as a UK village. (Presumably this comparison relies on a misplaced decimal point: a 2 megawatt power plant will feed around a thousand homes and there aren’t many UK villages with 10,000 houses.)
Exponential Progress mixes the picture-heavy approach of a glossy magazine (not quite so effective in black-and-white) with an attempt to cover a wide range of technologies that are likely to affect our future. But the breadth of subject matter is rather undermined by the mistakes, which is a shame because plenty of interesting questions are raised that would benefit from a better-grounded discussion.
What are the dangers of home CRISPR kits? Just where does Elon Musk’s decision to take the money he got selling PayPal to eBay and use it to buy Tesla and build SpaceX fall on Maslow’s hierarchy of needs and achieving self-actualisation? Are we prepared for the dangers of Artificial General Intelligence (AGI)?
But many readers will get a better definition of the problem in False Value, the latest volume of Ben Aaronovitch’s Rivers of London series, where protagonist Peter Grant handily explains that AGI is “the sort that was self aware enough to pass the Turing test and ask difficult philosophical questions before going ‘daisy-daisy’ and trying to wipe out humanity, while ordinary AI mainly tried to sell you books on Amazon”.
Sadly, Exponential Progress doesn’t have nearly so pithy a turn of phrase.
RECENT AND RELATED CONTENT
10 tech predictions that could mean huge changes ahead
Gartner sees ‘internet of behaviors,’ automation, AI, experiences key 2021 technologies
Singapore startup develops COVID-19 breath detection test
Intel sends AI to space in launch of a satellite the size of a cereal box
Fluree, the graph database with blockchain inside, goes open source
Read more book reviews
