Page 13 - Machine Learning for Subsurface Characterization

Preface xxv

6. Human beings have limited memory, cannot visualize data in high dimen-
sions, and have restricted cognitive resources for solving very complex
problems. ML tools help us handle complexity leading to improved human
productivity and efficiency. Notably, ML tools if designed properly can
remove human bias from decision-making.
7. Large firms are using ML tools to solve large-scale, high-visibility business
and engineering problems. Smaller firms should try to identify the
neglected, mundane tasks and deploy ML to solve them. The hype around
ML has made large firms to invest their energy on eye-catching,
news-worthy, marketable tasks. Moreover, there have been massive extrap-
olations of current ML trends and successes toward many exciting yet super-
ficial future scenarios. More useful applications of ML can only emerge
when we try to solve mundane and “boring” applications, which may never
get the limelight.

Concluding remarks
At the start, the field of genetics didn’t have any understanding or even theory of
DNA. Genetics in early days tried to answer simple, narrow tasks, such as “Why
some people have black hair?”. In the course of few decades, with the advance-
ments in biology, chemistry, microscopy, and computations, now, we can
sequence the whole human genome and understand physical basis of diseases
and traits. In the same vein, the field of AI is slowly marching toward the grand
vision of general intelligence, and ML/DL tools are few techniques helping us
progress the field of AI by harnessing the power of big data [10]. Once 3d
printing and virtual reality were in their hype phase. Both the technologies
are now coming out of the Trough of Disillusionment (Gartner Hype Cycle)
with real and useful applications. At the peak of hype, these technologies were
touted to accomplish grand tasks, for which they were not ready. AI/ML tech-
nologies are in the hype cycle but will soon come out of the hype much stronger
and more productive. While AI and its subsets are powerful tools capable of
shaping a wide range of industries and the way we live, they are not the ultimate
solution to the problems faced by us and our planet.
Siddharth Misra
Harold Vance Department of Petroleum Engineering,
Texas A&M University, College Station, TX, United States

References

[1] https://www.wired.com/insights/2015/02/myth-busting-artificial-intelligence/.
[2] https://builttoadapt.io/why-the-ai-hype-train-is-already-off-the-rails-and-why-im-over-ai-
already-e7314e972ef4.
[3] https://www.fico.com/blogs/analytics-optimization/hype-and-reality-in-machine-learning-
artificial-intelligence/.