May 7, 2020
Today on AI Australia, we have
the opportunity to talk to Professor Jon Whittle of Monash
University about the impacts - both good and bad - data science is
having on the world around us. As co-author of a series of
published and soon-to-be-published papers in the fields of software
development, ethics, and values, Jon is well placed to talk to us
today about the heightened risks and opportunities that the
development of data-science based systems brings to our
world.
About Professor Jon Whittle:
Professor Jon Whittle is the
Dean of the Faculty of Information Technology at Monash
University.
Jon is a world-renowned expert
in software engineering and human-computer interaction
(HCI), with a
particular interest in IT for social good. In software engineering,
his work has focused around model-driven development (MDD) and, in
particular, studies on industrial adoption
of MDD.
In HCI, he is interested in ways
to develop software systems that embed social values.
Jon's work is highly
interdisciplinary. As an example, he previously led a large
multidisciplinary project with ten academic disciplines looking at how
innovative digital technologies can support
social change. In 2019
Monash launched the Data Futures Institute, which we will find out
more about today.
Before joining Monash, Jon was a
Technical Area Lead at NASA Ames Research Center, where he led
research on software for space applications. He was also Head of
the School of
Computing and Communications at
Lancaster University, where he led eight
multi-institution,
multi-disciplinary research
projects. These projects focused on the role of IT in society
and included digital health
projects, sustainability projects and projects in digital
civics.
- 2:30
How is Monash busting the old model of disciplinary siloed schools
and turning toward a passion for multidisciplinary studies? How
does the intersection of fields lead to progress, especially in
software engineering and AI?
- 5:00
How do you go about raising awareness of these types of social,
ethical, and psychological aspects of interdisciplinary work and
studies with a deep technical kind of audience who are focused on
their tools and being the best they can in their particular arena?
What role do universities play in creating engineers who will
consider ethics and values in their software products?
- 8:00
What are values? Jon gives us a long and short answer that can
include everything from social responsibility to hedonism to
inclusion. What role do social scientists play in how we understand
values? Jon discusses
Schwartz’s Ten Universal Values, corporate values, and the difference between
ethics and values—as well as the ability for them to contradict one
another. How do values differ by culture, age group, and other
demographic factors?
- 14:00
Who gets dominance in a software application—who chooses the values
that underpin the software? How do we take the implicit aspect of
values in software and turn it into an explicit process? Jon
discusses the maturity scale of companies and their corporate
values and whether or not this impacts design
decisions.
- 19:00
Jon discusses the impact of corporate values on software
development and real-world cases of ethical issues that have arisen
due to software. This includes everything from parole re-offender
predictions, priority one-day shipping, and self-harm on
Instagram.
- 23:00
Are values at the root of algorithmic bias? Different groups
experience products and services differently, especially if the
data and ideas going into it are heavily biased. 80% of AI
professors are male—how does this influence the way systems are
designed? Are our programs today working to increase diversity in
AI and software development and, even when those fields are
diverse, what difference will it make? Jon proposes someone on the
team to specifically ask questions about values during the design
process.
- 29:00
Will we ever get to an empirical state where values are so
measurable that we could be alerted, automatically, of a breach?
Jon discusses the complexities of this process, but shares progress
that is already occurring on this front. Even without perfect
accuracy, we can see if things are getting better or
worse.
- 32:00
Is there a government role in policing the ethical use of software,
data, and AI? Jon shares his thoughts on a multi-faceted approach
to regulation.
- 35:00
How is Monash helping students prepare for a values-first data
environment? Jon discusses the “Bachelors of Applied Data Science”
multidisciplinary degree and the combination of “data plus x” in
education and the workforce.
- 38:00
Jon discusses examples where values went right and why we need
values built-in to software upfront, the way we do with security.
Jon also answers the questions, “Do robots have values?” with a
surprising standpoint.