This article serves as an introduction to a webinar that the author presented as part of AMC’s involvement in PDAC 2020. Click the button below to watch the webinar recording.
The Fourth Industrial Revolution, Industry 4; The Second Machine Age, The Knowledge Revolution, Life 3.0, Society 5.0, are terms now familiar to most people, but are the implications of this “Brave New World” understood? More specifically, what impact will they have for geologists and the way geology is taught and practised; or will geologists just fade away, a quaint practice associated with a primitive past?
The term “Fourth Industrial Revolution”, which is used in this article, was first defined by Klaus Schwab, the founder and executive chairman of the World Economic Forum (WEF), in 2015 and later was used as the title of his 2016 book The Fourth Industrial Revolution.
So, what is the Fourth Industrial Revolution and in what way could this impact geology? Simply put, the Fourth Industrial Revolution will integrate apparently disparate sciences and activities, mainly as a result of the advances made in computer processing speed, data storage and communications technology, but also from the integration of apparently disparate ontologies, resulting in new fields of study such as medical geology, geomicrobiology, and geobotany. However, how will the Fourth Industrial Revolution distinguish itself from the industrial revolutions that have preceded it (Table 1)?
Table 1: What is the Fourth Industrial Revolution?
The ability to store and analyze large amounts of data, in order to identify unrecognized patterns and relationships, is of interest to any earth scientist, especially when looking for new targets in previously explored terrains, or when applying analogue models to unexplored areas. The risk in this approach is to assume that by throwing reams of data at an algorithm, a useable answer will result. An axiomatic statement you may think. Regrettably the common belief appears to be that a smart algorithm, coupled with reams of data and unlimited processing power, can overcome the errors and omissions associated with poor quality data. Correlation is not causation, a fact known since data was analyzed by the first computers. So, to recall the mantra of the first computer programmers (dating back to the misty dawn of the computer age, 1957):
GIGO: Garbage in = Garbage out
Charles Babbage was more eloquent, in 1864, when confronted with the problem of data:
On two occasions I have been asked, “Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?” … I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question.
Regardless of the heritage, the concept remains as valid now as it was then:
Most geological data are derived from mundane, repetitive tasks. The classic example is core logging, vitally important to any exploration program (or operational mine). However, the task is usually delegated to relatively inexperienced geologists, due to the sheer volume and repetitive nature of the work. Various opinions have been bandied around in an attempt to quantify how much time a geologist spends logging core. Numbers vary, however a general consensus appears to be around 80%. When viewed in this context, it is not surprising that geologists fare badly in a study undertaken by the University of Oxford in 2013. In this study, Frey and Osborne estimated that the likelihood of a geologist being replaced by a robot/AI was 63%. Compare that to a mining engineer whose replacement likelihood was estimated to be 13%.
Which brings us back to data quality and reliability – how many of us have scratched our heads when trying to make sense of the data gathered by multiple geologists over multiple drilling programs, or even looked back at your own logs and wondered what the heck were you doing? Or the struggle to find that old core, only to be told it was discarded as part of a cost-cutting exercise?
Wouldn’t it be nice to have a data set that includes multiple parameters recorded in a consistent and reliable manner? The reality is that geologists are human, and we come with our own conceits, prejudices, cognitive biases and bad habits (including core logging) even when we are provided with standardized logging templates. In a paper by Lokier and Al Junaibi (2016), they found that, even with the use of a well-known, industry-wide, classification system (the Dunham Classification System), “ambiguities in the petrographic description of carbonates are widespread…. The most common causes of inconsistency are; errors in assessing the mode of support, mistakes in estimating the size and volume of grains within the lithology, and confusion as to how to classify lithologies in which more than one texture is present”.
So, if we start taking repetitive tasks away what is left for a geologist to do? This issue has been examined by the WEF in their Future of Jobs report, where they compare the skills that were required pre (2015) and post (2020) the Fourth Industrial Revolution (Table 2).
Table 2: Skills needed in the Fourth Industrial Revolution as determined by the WEF
The skills have been ranked in order of perceived importance. As we can see, the soft skills required in the Fourth Industrial Revolution are those associated with creativity and lateral thinking. As Jun Cowan points out “Real AI—the sort that can replace the sparse dot-connecting ability of trained Geologists—simply doesn’t exist”. However, the concept of IA (Intelligence Augmentation), is more likely the reality that we will see, for example geologists employing a combination of “intuitive deduction”, that deep and inscrutable, ineffable, skill that resulted in the postulation and discovery of Carlin–type gold deposits, Fipke and Blusson’s discovery of Canadian Diamonds, or Garnett’s discovery of Voisey’s Bay, in association with Big Data and Machine Learning.
With the advent of automatic core scanners that can also record, inter alia, X-ray computed tomography (CT) data, multiple data types can be accurately recorded and then interrogated. For example, core photographs, porosity data, color, lithology and geophysical data can be integrated, and these data then used in the development of a 3D geological model. As the data is recorded in three dimensions, the relationships between mineral species, phases and structures present in the core (and by extrapolation in the deposit) can be determined and modelled, and with the power AI’s like Microsoft’s Oxford, IBM’s Watson, Google’s DeepMind and Baidus’ Minwa, multiple models and scenarios can be generated.
As Pavel Abdur-Rahman (IBM Head of Data, IoT, AI and Advanced Analytics Practice) said: “we actually would like the geologists to do more geology and let machines do the data manipulation, information extraction and even some prediction”.
This is not the only area where geologists can expect to see material changes in the way we undertake our jobs, I doubt that we will fade away, but we will have to adapt and move with the times.
A brief note on computing power, data storage and communications technology:
A common example illustrating the advances made in computing is that of the Apollo Guidance Computer (AGC). In 1969, Eagle, the Apollo 11 lunar module landed Neil Armstrong and Edwin Aldrin on the moon. The final landing was controlled by the AGC, which had 4kB of RAM. Compare that to an iPhone XS that ships with 4gB of RAM and is used mainly to share cat memes. An alternative way of understanding how limited the AGC is to think that if you were to use it to mine Bitcoin, it would take the AGC over a quintillion year (a billion billion) to mine a single block.
The ability to access limitless amounts of knowledge wherever you are, is coming courtesy of the ever-expanding World Wide Web (the proposed Russian and Chinese UDIs remain to be seen). According to Wikipedia, the largest paper encyclopedia ever produced was the Yongle Encyclopedia (more correctly a leishu, a Book of Knowledge) completed in 1407 at the order of the Yongle Emperor of the Ming dynasty comprising 1,095 printed volumes. Compare that to what is instantly accessible at Wikipedia (as of 24 January 2020, there are 6,000,824 articles in the English Wikipedia alone, with more than 270 languages having their own Wiki. (Wikipedia is a portmanteau word, derived from “wiki”, the Hawaiian word for “quick” and “encyclopedia”),
And finally, communications technology. In the 1983 film WarGames, Matthew Broderick almost sparks a nuclear war when he hacks WOPR (War Operation Plan Response, an early AI) using an acoustic coupler linked to a dial-up modem (and for those that forget, and for those of us that only ever knew it as a “modem”, modem is another portmanteau word modulator – demodulator). Typical transmission rates were 150 bits per second (bps, baud), compare that to a 4G network, where download speeds can be as high as 60Mbps allowing teenagers (and others) to MMOG for a Fortnight.
Lokier, S.W and Al Junaibi, M, (2016), The petrographic description of carbonate facies: are we all
speaking the same language? Sedimentology (2016) 63, 1843–1885