
Outline
Ø Application of AI-Ethics in Engineering
Ø Example of AI-Ethics in Production from Shale Wells
Ø Conclusions
Application of AI-Ethics in Engineering
Bias (including major assumptions, interpretations, and simplifications) from traditional engineers can be included in the engineering application of AI. This is usually done through the generation of data from mathematical equations and combining it with actual field measurements (actual physics-based data) and then using this combined set of data to model the physics using AI and Machine Learning algorithms. In many cases, such an approach is called "Hybrid Models". In the context of engineering application of Artificial Intelligence and Machine Learning such models is the determination of lack of realistic and scientific understanding of AI and Machine Learning.
The Ethics of Artificial Intelligence is important to engineers and scientists that have become enthusiasts to use this technology for solving engineering-related problems. While in the past several years AI-Ethics has become an important topic in the non-engineering application of Artificial Intelligence and Machine Learning, now it is just as important in the engineering application of this technology. A specific example of AI-Ethics in the engineering application of AI and Machine Learning is presented in the next section showing how guesswork, assumptions, interpretations, and simplifications can help traditional engineers to use AI and Machine Learning algorithms to generate an unrealistic and highly biased predictive model. This usually happens when they have tried but have not been successful to use facts and field measurements.
It seems that the reasons behind the inclusion of such biases in the engineering application of AI have much to do with the lack of scientific understanding of how Artificial Intelligence must be used to model physical phenomena. Currently, some individuals and companies that claim the use engineering application of this technology are including a large amount of human biases so that they can solve problems using AI after they fail to build an AI-based model that does not include human biases. Human biases in engineering have much to do with how mathematical equations are built to solve physics-based problems.

The major contribution of AI and Machine Learning to engineering problem solving is modeling the physical phenomena based on actual measured data that would be the main core behind the avoidance of biases, assumptions, interpretations, and preconceived notions about physics. Since traditional techniques for modeling of the physical phenomena is through mathematical equations, it usually includes assumptions, and sometimes biases. This is very true when the physical phenomena that is being modeled cannot be seen, looked at, or even touched such as Petroleum Engineering that is a good example of such a situation since the produced hydrocarbon is a deeply underground fluid. The same is true about any other engineering discipline when the mathematical equations that include assumptions, interpretations, and simplification are used to model the physical phenomena.
Reservoir engineering, reservoir modeling, and reservoir management overwhelmingly contribute to the majority of the income of the operating and service companies in the oil and gas industry. This shows why reservoir modeling is a very important technology in the petroleum industry. It is a fact that modeling fluid flow in hydrocarbon reservoirs includes serious amounts of assumptions, interpretations, and simplifications since they are hundreds or thousands of feet below the surface. This means that it has been impossible to actually observe, touch, or realistically test anything that takes place in a hydrocarbon reservoir.
It is important to note that some part of the reservoir rock is usually brought to the surface and it is tested and analyzed in the laboratory to help scientists and engineers develop an understanding of the complexity of physics of fluid flow in the porous media, deep underground. However, realistic facts about such analyses must not be overlooked. While hydrocarbon reservoirs have approximate volumes of hundreds of millions to tens of billions of cubic feet, the part of the hydrocarbon reservoirs rock that is brought to the laboratory for observation, and testing are usually less than a few cubic feet. Furthermore, it is a well-known fact that hydrocarbon reservoirs are highly heterogeneous which means what is analyzed in the laboratory over a few square inches of rock is not realistically representing what happens throughout the entire reservoir rock.
Therefore, while laboratory core analysis is an important and useful process of understanding the fluid flow in the hydrocarbon reservoir, they cannot realistically represent all the details and the heterogeneity that happen throughout the tens of billions of cubic feet throughout the hydrocarbon reservoir that is hundreds or thousands of feet under the ground. This clarifies the existence of assumptions, interpretations, and simplifications in the mathematical equations that are used to model the fluid flow in porous media. Furthermore, when the hydrocarbon reservoir is unconventional such as shale plays that currently is the main source of hydrocarbon production in the United States, the problem mentioned above becomes orders of magnitude more complex and even more important.
Example of AI-Ethics in Production from Shale Wells
The number of assumptions, interpretations, simplifications, preconceived notions, and biases in modeling the physics of completion, hydraulic fracturing, and fluid flow in shale plays is so extensive that make the mathematical modeling for completion and production optimization from shale wells completely unrealistic, useless, and full of bias. This is due to the fact that historical details of understanding the physics of fluid flow in hydrocarbon reservoirs are mainly applicable to conventional plays such as sandstones and carbonates. This technology that has been developed for conventional reservoirs has been extrapolated to unconventional reservoirs starting a decade ago.
It is a clear fact that the mathematical equations that are used to model the physics of hydrocarbon production from shale wells are overwhelmed by assumptions and hardly have anything to do with facts and realities since the main essence of this technology is mainly applicable to the conventional reservoir and not to the unconventional reservoir. It is hard to find any real scientists and professional engineers (including those that have developed and are using these techniques) to claim that the current version of mathematical modeling of hydraulic fracturing of shale wells has anything to do with reality.
These facts prove that using AI and Machine Learning for the development of so-called "hybrid models" is full of assumptions, interpretations, and biases and has very little to do with the reality of engineering application of Artificial Intelligence and Machine Learning. When such mathematical equations are used to generate data and then combine such data with actual field developments in order to build so-called "hybrid models", such models can be forced to generate the type of outputs and results that is pre-determined by those that develop it. It removes the actual and real characteristics of Machine Learning Algorithms that are capable of modeling physics based on reality rather than based on guesswork and biases. This is a good example of how AI-Ethics must be addressed in the engineering application of this technology.
It is a well-known fact that when hydraulic fracturing is performed on unconventional reservoirs that are naturally fractured (such as shale) the results are quite different from what happens when hydraulic fracturing is performed on conventional reservoirs (sandstones). In shale, due to the existence of the complex natural fractures, hydraulic fracture creates a "network of fractures" (as shown in Figure 1 and Figure 2), not an elliptical hydraulic fractur (as shown in Figure 3 for conventional reservoirs).
As shown in Figure 1 and Figure 2, when liquid (water) is injected in an unconventional reservoir for hydraulic fracturing purposes, prior to the injection of the proppant, it starts to frac the formation. As the formation starts to fracture, the continuation of the fracture will go through the least resistant pathways in the rock. In a naturally fractured reservoir, the least resistant pathway is the network of the natural fractures while the actual fabric of the rock (that has not been naturally fractured millions of years ago) has more resistance. Therefore, hydraulic fracturing of unconventional resources such as shale that are naturally fractured reservoirs creates highly complex networks of natural fractures that cannot be modeled in detail. This is due to the fact that the shape, characteristics, and details of the natural fracture of the rock (shale) cannot be observed or measured throughout the reservoir. The highly complex shape of the hydraulic fracture network in the unconventional reservoir is a function of heterogeneity and natural fracture network.


When the model development of hydraulic fracturing was performed more than 50 years ago the shape of the hydraulic fracture in the conventional reservoir was modeled using an elliptical shape as shown in Figure 3. This traditional hydraulic fracture model includes four specific characteristics that allow it to be modeled using mathematical equations. These four specific characteristics are (a) fracture half-length, (b) fracture height, © fracture width, and (d) fracture conductivity. Comparing the shapes of the hydraulic fractures that are demonstrated in Figure 1 and Figure 2 versus the shapes of the hydraulic fracture that is shown in Figure 3 makes it quite clear how different the actual shape of hydraulic fracture is between unconventional versus conventional reservoirs.
When the actual hydraulic fractures look like what is shown in Figure 1 and Figure 2, does it make any sense or does it have anything to do with reality, to model it using the shape that is shown in Figure 3? The answer to this question must be quite clear. This is a good example of how assumptions, interpretations, preconceived notions, simplifications, and biases that are included in the mathematical equations (that are used to model the physics of fluid flow in porous media) are included in "hybrid models" that combine them with real field measurements.

Conclusions
The Ethics of Artificial Intelligence has proven to be an important issue when AI-based models are used in decision making. AI-Ethics can expose biases that may have been included in the AI applications of Engineering and Non-engineering problem-solving. Several research, some of which were referenced in this article, had shown how biases such as racism and sexism have been included in the AI-based models that have been exposed through AI-Ethics. This article demonstrated how assumptions, interpretations, and biases developed by traditional engineers can be included in the engineering application of Artificial Intelligence that is referred to as AI-Ethics in Engineering.
References
· Crawford, K. and Paglen, T., Excavating AI, The Politics of Images in Machine Learning Training Sets, September 2019 – https://www.excavating.ai/
· Dastin, J., Amazon scraps secret AI recruiting tool that showed bias against women, Reuters – San Francisco. October 2018.
· Dave, P., Exclusive Google Searches for New Measure of Skin Tones to Curb Bias in Products. June 2021. https://www.reuters.com/business/sustainable-business/exclusive-google-searches-new-measure-skin-tones-curb-bias-products-2021-06-18/?utm_medium=techboard.fri.media.20210618&utm_source=email&utm_content=&utm_campaign=campaign
· Doshi, T., Introducing the Inclusive Images Competition, September 6, 2018, Product Manager, Google AI – Google AI Blog
https://ai.googleblog.com/2018/09/introducing-inclusive-images-competition.html
· Hao, K., Making face recognition less biased doesn’t make it less scary. MIT Technology Review. Tech Policy – Ai Ethics.
· Hao, K., This is how AI Bias really Happens – and Why it’s so Hard to Fix. MIT Technology Review. Tech Policy – AI Ethics.
· Lim, H., 7 Types of Data Bias in Machine Learning, July 20, 2020
https://lionbridge.ai/articles/7-types-of-data-bias-in-machine-learning/
· Terence, S., Real-life Examples of Discriminating Artificial Intelligence. Real-life examples of AI algorithms demonstrating bias and prejudice. June 2020.