Ethics in Artificial Intelligence (AI)/Machine learning (ML) and Deep Learning - Part 17
by
RGS Asthana
Life Senior, IEEE
Summary
is ethics a moral Issue or an emotional issue? In fact, it is ultimately both. We have discussed this aspect with respect to AI/ML in this paper.
Prerequisite
Read articles [1] to [22]
Keywords
Machine Learning (ML) Tools, ML.Net, Artificial Intelligence (AI), Neural Networks, Internet of Things (IoT) and Deep Mind.
Prelude
We need to read articles from [1] to [20] to know about AI/ML and also neural nets which are the heart of ‘Deep Learning’. In the year 2018, we have seen the progress of data adoption [23] in business and in the year 2019, we would see a continuation of this trend leading to the conversion of businesses worth the name to turn into a data-driven business. If businesses do become data-driven then one can appreciate the value of data, particularly, corporate data. X But the recent unauthorized use of personal and social data by Facebook and Cambridge Analytica [26] has once again brought ethical aspects in frontend [24] leading to the feeling that there should be some ethics in the corporate world on the use of the data. in short, we need to worry about both the privacy and security of the corporate data.
AI/ML depends on big data for various applications, e.g self-driving vehicles, trading, clinical decision support systems and data mining as the accuracy of the AI/ML algorithm is based on the amount of data it is trained on. This is, particularly, true for systems based on Deep Learning technologies. Both India and China have large populations but they have digitized their economies. India has gone through demonetization in 2016 and has an integrated goods and services tax (GST). China has implemented GST in the year 2012 and final corrections were made in the year 2016.
recent news is that both Tesla and Uber have slowed down their efforts to develop autonomous vehicles because of accidents. Both need more data to improve their vehicle’s performance on the road. This all brings us to a very pertinent question, How do we ensure the ethical and responsible use of AI? This is where we feel that standards can play a great role and defining Global ethics in the use of available data can solve this problem to a very good extent.
What is ethics?
As per dictionary definition of Ethics [28] means:
”moral principles that govern a person's behavior”
E.g. think of abortion and euthanasia, they are moral as well as the emotional issue. Ethics, as said earlier is a system of moral principles.[27]. They affect how people make decisions and lead their lives.
Ethics is, in fact, moral philosophy and is concerned with what is good for people and society.
Ethics covers the following dilemmas [27]:
What use is ethics [27]?
Ethics need to affect the way one behaves.
But if one does often behave irrationally. One solution to this problem could be that he/she may follow his/her 'gut instinct' even when his/her head suggests a different course of action.
However, ethics provides a basis for thinking about moral issues e.g.
Big data, AI/ML and Ethic
Are Software Developers, Software Engineers, Data Scientists, Data Engineers, etc, will decide what the impact of these technologies are and whether to replace or augment humans in society? The answer to this question is most certainly ‘no.but right people need to take the lead.
The need for ethics is more urgent as with Big Data, AI and Intelligent systems do become complex and more effective tools in the hands of a variety of stakeholders, including politicians.
Some AI/ML applications may raise new ethical and legal questions, for example, related to liability or potentially biased decision-making.
Is it OK if the business’s self regulates themselves? All big companies worth the name [30-34] have AI guidelines which include clauses such as social benefit, AI developed by them will be fair to everyone, commitment to conduct intellectually rigorous and evidence-based interdisciplinary research and also commitment to develop AI which is reliable and safe i.e. it should be understandable by everyone and should be secure and respect privacy as well as algorithmic accountability.
A new AI research institute is launched recently at New York University. It is dedicated to the study the social implications of AI/ML and algorithmic accountability.[35]. The beauty is that it’s the first of its kind and also launched by all women only.
Deep Learning and Ethics
In terms of ethical questions, Amazon’s replaced hiring tool [37], based on AI, was biased towards hiring men. This demonstrates that AI is only as good as the data put into it.
‘AI is only as good as the data put into it’
–- BRIAN MCDERMOTT (å full-stack developer at Allstate)
A self-driving-vehicle [36] may have to take a decision which of people to knock down a person crossing the road or the person in the car. Decisions are hard regardless of who’s making the decision. Even as flawed as humans’ morals are, what another baseline could we use to code machines? There is no other baseline to be used.
‘If we employ artificial neural networks, the network relearns from existing predictions. This is similar to how the human brain works’
— ANGELINA VILLIKUDATHIL, ( A researcher at Ulster University)
In Fact, Deep Learning is an unsupervised system and is neural net based. It, at present, does not give the reason why it has arrived at a particular decision. Deep learning learns relationships with the data. It trains a neural network and finds a level of patterns we can’t see with the human eye as it has bandwidth limitations.
Look at a situation where we define a rule that you can’t allow a computer to make a decision that you can’t explain as a human This is the only way computers can follow Ethics.
Amazon’s experience to automate the hiring process did serve as a deterrent to the companies including Hilton Worldwide Holdings Inc [38] and Goldman Sachs Group Inc. These companies looked at the alternate ways to automate portions of the hiring process. E.g. (1) Hilton Worldwide Holding Inc. provided a platform to allow for convenient video interviews, it also included artificial intelligence (AI) features that evaluate candidate’s communication skills. For customer service roles, Hilton’s recruiters can even use simulations to test how a candidate would react to an angry customer. This approach has cut the recruitment process from 42 days to just 5 days, without compromising the quality of candidates selected; (2) Google [38] implemented a “Rule of Four” standard i.e. a prospective candidate will be interviewed maximum by 4 recruiters. Using a maximum of four interviews, the company cut time to hire by about two weeks—making the process less stressful for candidates and also savings for the company. (3) UK’s NHS Hospitals Foundation Trust [40], provided specialized care for over two million patients every year. In 2015, the time to hire was 18-24 months and it has reduced time by more than half (down to eight to 10 weeks). The 127-step process was stripped back to 53 helpful steps.
Way Forward
With Global use of AI/ML, it will become really difficult to attribute responsibility for decisions to any single or a group in particular. If mistakes are made which cause harm, who should bear the risk? When complex ML systems are used in an organization to arrive at key decisions, it may be difficult to unpick the causes behind a specific course of action. Clear explanations for machine reasoning are necessary to determine accountability. This is, particularly, applicable to Deep Learning Algorithms.
The BS 8611 standard developed by British Standard Institute [25] on the “ Robots and robotic devices: Guide to the ethical design and application of robots and robotic systems: "Robots should not be designed solely or primarily to kill or harm humans. Humans, not robots, are the responsible agents; it should be possible to find out who is responsible for the design of any robot and its behavior.”
Potential and accuracy of AI/ML System are in the access methods adopted to get access to large datasets. What happens when an AI system is trained on one data set and wishes to use a new data set for testing? I think it should be an ethical use of the other dataset.
However, Amazon, Google, Facebook, IBM, and Microsoft have established a non-profit partnership to formulate best practices on AI/ML technologies. Hopefully, this will pave the way for the ethical use of AI/ML technology.
References
[1] Progress and Perils of Artificial Intelligence (AI)
[2] Invited Chapter 6 - Evolutionary Algorithms and Neural Networks, Pages 111-136, R.G.S. Asthana in the book, Soft Computing and Intelligent Systems (Theory and Applications), Academic Press Series in Engineering, Edited by Naresh K. Sinha, Madan M. Gupta and L.A. Zadeh ISBN: 978-0-12-646490-0
[3] Future 2030 by Dr RGS Asthana
[4] Machine Learning (ML) and Artificial Intelligence (AI) – Part 1, by Dr. RGS Asthana, Senior Member IEEE
[5] Machine Learning (ML) and Artificial Intelligence (AI) – Part Two, by Dr. RGS Asthana, Senior Member IEEE3
[6] Machine Learning (ML) and Artificial Intelligence (AI): Cognitive Services and Robotics – Part Three by Dr. RGS Asthana, Senior Member IEEE
[7] Machine Learning (ML) and Artificial Intelligence (AI): Big Data and 3 D Printing – Part four by Dr. RGS Asthana, Senior Member, IEEE.
[8] Machine Learning (ML) and Artificial Intelligence (AI): Drones and Self-driving Cars– Part Five by, Dr. RGS Asthana, Senior Member IEEE
[9] Machine Learning (ML) and Artificial Intelligence (AI): Healthcare– Part Six by, Dr. RGS Asthana, Senior Member IEEE
[10] Machine Learning (ML) and Artificial Intelligence (AI): Will AI/ML intelligence surpass humans? Part Seven by Dr. RGS Asthana, Senior Member IEEE
[11] Machine Learning (ML) and Artificial Intelligence (AI): Impact of AI/ML in Healthcare: Part-Eight by Dr. RGS Asthana, Senior Member IEEE
[12] Machine Learning (ML) and Artificial Intelligence (AI): Big data & Data Science (DS) and their importance: Part-Nine by Dr. RGS Asthana, Senior Member IEEE
[13] Machine Learning (ML) and Artificial Intelligence (AI): Super-Intelligence - Are we afraid? Part-ten; by Dr. RGS Asthana, Senior Member IEEE.
[14] Machine Learning (ML) and Artificial Intelligence (AI): ML Algorithms: Part- Eleven
[15] Machine Learning (ML) and Artificial Intelligence (AI): Prominent ML & AI applications including those on Mobile devices: Part - Twelve
[16] Robotics advances with Machine Learning (ML) and Artificial Intelligence (AI) and its impact on healthcare Part - 13
[17] Deep mind website
[18] IBM Watson Website
[19] Internet of Things (IoT)
[21] Comparison of key frameworks: Machine Learning (ML) and Artificial Intelligence (AI) Part - 15
[22] Impact on Drug development and design due to latest advances in Machine Learning (ML) and Artificial Intelligence (AI) in healthcare Part - 14
[23] A 2019 Forecast for Data-Driven Business: From AI to Ethics
[24] We know ethics should inform AI. But which ethics?
[25] BSI - BS 8611 Robots and robotic devices Guide to the ethical design and application of robots and robotic systems
[26] Facebook and Cambridge Analytica: What You Need to Know as Fallout Widens
[27] Ethics: a general introduction
[28] Dictionary definition of ethics
[29] Big Data and AI– Ethical and Societal implications
[30] DeepMind Ethics & Society Principles
[31] Microsoft AI principles
[32] Intel and AI
[33] AI at Google: our principles
[34] Siia Issue Brief Ethical Principles for Artificial Intelligence and Data Analytic
[35] New AI Research Institute launches at NYU
[36] What are the ethical implications of deep learning?
[37] Amazon scraps secret AI recruiting tool that showed bias against women
[38] How Hilton, Google, and More Have Dramatically Reduced Their Time to Hire
[40] NHS authorities and trusts
|
Happiness V 3.0
Wednesday, 30 January 2019
Ethics in Artificial Intelligence (AI)/Machine learning (ML) and Deep Learning - Part 17 by RGS Asthana Life Senior, IEEE
Subscribe to:
Posts (Atom)