There’s an interesting debate going on the in AI community. It has actually been happening for a while, it’s just that recently it has become more public and more personal.
Not exactly the old nature vs. nurture discussion but
something similar. The essential question is: What is intelligence and how do
AI-agents become intelligent (more intelligent)? Let’s simplify the debate and
make it about two people: Rich Sutton and Gary Marcus.
Rich Sutton is the leading proponent of reinforcement
learning (the trial, error, and reward based learning often associated with
Gary Marcus is a cognitive scientist who has often called for
the return of symbolic AI (e.g. GOFAI) and is currently advancing what he calls
In a nutshell:
Sutton, The Bitter Lesson (2019), believes that intelligence is computation and that all we need to do is leverage computational scale and intelligent agents/systems (actions) will emerge. All the human knowledge, building in what AI folks call “priors”, is a “distraction” – not worth the effort. General systems always win.
Score one for nurture.
Marcus, The Next
Decade in AI: Four Steps Towards Robust Artificial Intelligence (2020), does
not dispute that computation has a role in intelligence, he just doesn’t think it’s
sufficient or even efficient. People learn, in part, because they have “priors”
– innate understand and knowledge. We use that as building blocks to create
more knowledge as we experience new situations. Why not, says Marcus, use those
priors in AI to speed understanding and knowledge creation?
Score one for nature and
From my perspective, both approaches are interesting and, frankly, valid (in their own way). The difference for me is the outcome. From my bias and interest in machine information behaviour, the agent resulting from the strategy of Sutton or Marcus will behave differently. And that’s OK. I think. Grin.
Is machine learning an environmental hazard or an
The answer is, apparently, “yes”.
Recently a number of papers have focused attention on machine
learning and climate change. Interesting findings.
Climate Change with Machine Learning” (Rolnick et al. 2019) is a manifesto
published in advance of the NeurIPS conference. This extensive and detailed
report outlines many ways in which applying ML can have a positive impact on addressing
significant aspects of climate change. In summary:
“ML can enable automatic monitoring through remote sensing (e.g. by pinpointing deforestation, gathering data on buildings, and assessing damage after disasters). It can accelerate the process of scientific discovery (e.g. by suggesting new materials for batteries, construction, and carbon capture). ML can optimize systems to improve efficiency (e.g. by consolidating freight, designing carbon markets, and reducing food waste). And it can accelerate computationally expensive physical simulations through hybrid modeling (e.g. climate models and energy scheduling models).”
“The estimated 2020 global footprint [of the tech industry] is comparable to that of the aviation industry, and larger than that of Japan, which is the fifth biggest polluter in the world. Data centers will make up 45% of this footprint (up from 33% in 2010) and network infrastructure 24%.”
They conclude that overall, “we see little action to curb
emissions, with the tech industry playing a significant role in the problem.”
While the Rolnick et al. report illustrates that applying ML
to environmental challenges has been and will continue to be productive, the
story is a bit different when looking at the environment challenges of training
the ML models to do this very work.
Strubell et al., “Energy
and Policy Considerations for Deep Learning in NLP” (2019), estimate that “training BERT [a widely used NLP model]
on GPU is roughly equivalent to a trans-American flight.” The authors of
“Green AI” (Schwartz et al.,
2019) note that the amount of compute required to train a model has increased
600,000 times (!) since 2013. More and more data, millions of parameters, and
hundreds of GPUs. And it’s getting
worse. They advocate “making efficiency an evaluation criterion for
research alongside accuracy and related measures. In addition, we propose
reporting the financial cost or “price tag” of developing, training, and
running models to provide baselines for the investigation of increasingly
Whatever the directions taken, the ML community, and the tech
industry more generally, are going to have to take their environmental impact much
more seriously. The role of environmental solution is possible but not at the
increased expense of environmental hazard.
Periodically the AI field has entered an “AI Winter” where the dominant paradigm seems to have run its course and researchers look for new options.
Are we entering another AI Winter?
Three recent books suggest not so much renewed stormy weather as a need to broaden perspectives … some looking backward, some merely looking around.
The basic questions raised are simple: Is Deep Learning (the state of the art in machine learning) sufficient? Is it the path to towards more intelligent machines (even AGI – artificial general intelligence).
Russell is widely known as the co-author of Artificial Intelligence: A Modern Approach (3rd ed. 2009), the definitive textbook in the field. In the past few years he has been exploring the concept of “beneficial AI” and this book further articulates that concept.
“The history of AI has been driven by a single mantra: ‘The more intelligent the better.’ I am convinced that this is a mistake.”
Russell. Human Compatible (2019)
Russell’s concern is that the current path of increasing AI autonomy fueled by more data, opaque algorithms, and enhanced computing will lead to a loss of control by humans. Not as bleak as Bostrom’s Superintelligence (2014), Russell’s solution is a design concept: make intelligent systems defer to human preferences.
Russell has three guiding principles:
The machine’s only objective is to maximize the realization of human preferences.
The machine is initially uncertain about what those preferences are.
The ultimate source of information about human preferences is human behavior.
Putting humans at the center of intelligent machines seems reasonable and certainly desirable. But will it be effective and advance AI?
The concern of Marcus (a long standing and vocal critique of Deep Learning) and Davis is related to Russell’s but the focus is different: not a control problem but a myopic problem – AI “doesn’t know what its talking about”; it doesn’t actually “understand” anything.
“The cure for risky AI is better AI, and the royal road to better AI is through AI that genuinely understands the world.” p. 199
Marcus & Davis. Re-Booting AI (2019)
And the way to understand the world is through “common sense”. In part this looks back to the symbolic (logic) representations of GOFAI (“Good Old Fashioned AI”) and it part it is about training AI about “time, space, causality, basic knowledge of physical objects and their interactions, basic knowledge of humans and their interactions.” Getting there requires us to train AI like children learn (an observation Turing made in 1950).
Smith picks up the issue of “understanding the world” and argues that AI must be “in the world” in a more visceral way – “deferring” to the world (reality) as we do. Two key concepts standout: judgment and ontology.
Judgment: Smith makes the distinction between “reckoning” (which most machine learning systems accomplish – calculation and prediction) and “judgment” which he views as the essence of intelligence and the missing component in AI.
Ontology: Smith contends that machine learning has “broken ontology.” It has given us a view of the world as more “ineffably dense” than we have ever perceived. The complexity and richness of the world require us to conceptualize the world differently.
The arguments about judgment and ontology converge in a discussion about knowledge presentation and point the way for machine learning to transcend its current limitations:
“If we are going to build a system that is itself genuinely intelligent, that knows what it is talking about, we have to build one this is itself deferential – that itself submits to the world it inhabits , and does not merely behave in ways that accord with our human deference.”
Smith. The Promise of AI (2019)
This book celebrates the power of machine learning while lamenting its shortcomings. However:
“I see no principled reason why systems capable of genuine judgment might not someday be synthesized – or anyway may not develop out of synthesized origins.”
The Collaboratory at the Ryerson University Library is starting a journal club to explore AI/ML. With this club, Jae Duk Seo, a graduate student at Ryerson, and I want to encourage a multidisciplinary perspective on AI/ML.
Too often the various disciplinary communities interested in AI/ML (computer science, engineering, mathematics, social sciences, humanities, etc.) talk among themselves …. to the detriment of an inclusive discussion.
This journal club will select topics and associated reports or papers that implicate a wide variety of disciplines (e.g. deepfakes, facial recognition, green AI, diversity, ethics, disinformation). Participants will bring their own readings of the issues and allow the group to explore the topic in ways that will open up new insights.
The hope is that exploring both common ground and different perspectives among the disciplines will create a challenging interchange for all. We want to encourage an informed discussion that incorporates all the aspects of AI/ML – technical, social, economic, and political.
Initially this group will meet in person; hopefully in the future we can accommodate both in-person and online.
At the core of machine learning are training datasets. These
collections, the most common are images, have labels (metadata) describing their
contents and are used by an algorithm to learn how to classify them. A portion
of the dataset is reserved for validation – testing the learned model with new,
previously unseen, data. If all goes well, the model is then ready to classify
entirely new data from the real world.
There are many such datasets and they are used repeatedly by
AI researchers and developers to build their models.
And therein lies the problem.
Issues with datasets (e.g. lack of organizing principles, biased
coding, poor metadata, and little or no quality control) result in models
trained with those problems and reflecting this in their operation.
While over reliance on common datasets has long been a concern (see Holte, “Very simple classification rules perform well on most commonly used datasets”, Machine Learning, 1993), the issue has received widespread attention because of the work of Kate Crawford and Trevor Paglen. Their report, Excavating AI: The Politics of Images in Machine Learning Training Sets, and their online demonstration tool, ImageNet Roulette (no longer available as of September 27th), identified extraordinary bias, misidentification, racism, and homophobia. Frankly, it will shock you.
Calling their work the “archeology of datasets”, Crawford
and Paglen uncovered what is well known to the LIS field: all classification is
political, social, and contextual. In essence, any classification system is wrong
and biased even if it is useful (see Bowker & Star, Sorting Things Out,
From an LIS perspective, how is ImageNet constructed? What
is the epistemological basis, the controlled taxonomy, and the subclasses? Who added
the metadata, under what conditions, and with what training and oversight?
ImageNet was crowdsourced using Amazon’s Mechanical Turk. Once
again, therein lies the problem.
While ImageNet did use the WordNet taxonomy to control
classifications, it is not clear how effectively this was managed. The results
uncovered by Crawford and Paglen suggest not very much. This year many training
datasets were taken offline or made unavailable, and many were severely culled
(ImageNet will remove 600,000
images). However, these datasets are important; ML relies on them.
Bottom line: the
LIS field has extensive expertise and practical experience in creating and
managing classification systems and the requisite metadata. We are good at
this, we know the pitfalls, and it is clear and compelling opportunity for LIS
researchers and practitioners to be centrally involved in the creation of ML
Who’s staffing the reference desk or the library chatlines? These days, or in the near future, it might be Google Assistant, Alexa, Cortana, or Siri. Library users may increasingly turn to virtual or personal assistants before they interact with specific library services. And why not, they appear to be getting quite good.
In 2018 Perficient Digital tested Alexa, Cortana, Google Assistant and Siri with nearly 5,000 questions :
BTW, they also tested which assistant was the funniest by tracking the jokes they made in response to some questions. “What is the meaning of life?” Siri: “All evidence to date suggests it’s chocolate.”
Results like this intrigued Amanda Wheatley and Sandy Hervieux of the McGill University Library. As a result, they initiated a multi-phase research project to explore the awareness of AI among libraries and librarians, their use of this technology, and what their expectations are for the future.
They believe AI will “change the nature of our work but won’t take our jobs.” AI will not displace librarians and library staff but operate as “an immersive environment where we coexist.” From their perspective “AI is not one thing” but an array of options and opportunities to be used in thoughtful ways. However, it is time to be proactive not reactive; we should lead in the use of this technology not be used by it.
Phase 1 (completed): an environmental scan of libraries and their use of AI as indicated in strategy plans or other documentation. The result? Not too much happening. This could be a lack of funds for technology innovation or it might be a concern about the nature of the technology.
Phase 2 (in process): a broad survey of libraries and librarians to assess their awareness and expectations of AI. That survey is currently live. The deadline for responses is September 6, 2019. You are encouraged to participate!
Phase 3 (in process): testing various devices with sample reference questions. The first test pitted Google against Siri with Google a clear winner. It responded by summarizing information, presenting relevant graphs and charts, and providing credible research materials … “it was terrifying!”. They are now starting to work with the Alexa Skills Kit to teach Alexa new library skills.
Phase 4 (planned): an AI experience in the McGill libraries to give the community a hands-on opportunity to explore the technology.
If you want more information about their work, visit their guide to the project or contact them via email: Amanda Wheatley (firstname.lastname@example.org) or Sandy Hervieux (email@example.com).
Amanda and Sandy are editing a book for ACRL on the use of AI in libraries. A call for chapters will go out in the fall.
[UPDATE: the call for chapters for this book is out. Deadline for proposals is November 17.]
Lots of interesting research to follow. Looking forward to hearing about their progress.
While AI text mining is not new, this article presents a new development that has important implications for research libraries:
Tshitoyan, V., Dagdelen, J., Weston, L., Dunn, A., Rong, Z., Kononova, O., … Jain, A. (2019). Unsupervised word embeddings capture latent knowledge from materials science literature. Nature, 571(7763), 95–100. https://doi.org/10.1038/s41586-019-1335-8
Of course, it’s from Nature; it’s behind a paywall. Sigh. Hopefully you are able to obtain a copy.
Using unsupervised methods of text mining in the area of materials science, the authors have demonstrated “that latent knowledge regarding future discoveries is to a large extent embedded in past publications.” The discoveries of the future were evident in the literature of past.
Using current and past literature, these approaches “have the potential to unlock latent knowledge not directly accessible to human scientists.”
“Such language-based inference methods can become an entirely new field of research at the intersection between natural language processing and science, going beyond simply extracting entities and numerical values from text and leveraging the collective associations present in the research literature.”
Interestingly, this possibility was explored much earlier during the formative years of MEDLINE albeit with less sophisticated tools:
The Tshitoyan et al. research is an exciting development using ML approaches that should become standard tools for research libraries. It is well worth your consideration. It is also, therefore, a concern that this work goes on without any involvement from libraries or those with LIS expertise.
Beta Writer algorithmically categorized and summarized more than 150 key research publications selected from over 1,000 published from 2016 to 2018. I’m no expert on lithium-ion batteries so others will have to weigh in on whether this is a credible book . However, a book that synthesizes and summarizes a large and complex corpus of current research literature is a valuable contribution.
The process of the book, a combination of various “off the shelf” natural language processing (NLP) tools, preprocesses the documents to address various linguistic and semantic normalizations, clusters documents by content similarity (i.e. chapters and sections of the book), generates abstracts, summaries, introductions, and conclusions, and outputs XML as a final manuscript. And it does so in a manner that is sensitive to copyright infringements. The details are outlined in a human written Preface (Henning Schoenenberger, Christian Chiarcos, and Niko Schenk) and provide an interesting comparison to current cataloguing and metadata processes and theories.
In an interview published in The Scholarly Kitchen, Schoenenberger was clear that the intent is “to initiate a public debate on the opportunities, implications and potential risks of machine-generated content in scholarly publishing.” This book is far from perfect and Springer acknowledges that. Commendably, Springer has gone to great lengths to document their process, discuss alternative strategies, identify weaknesses and outright failures, and to encourage critical commentary.
We foresee that in future there will be a wide range of options to create content – from entirely human-created content, a variety of blended man-machine text generation to entirely machine-generated text.
Henning Schoenenberger, Director Product Data & Metadata Management at Springer Nature
Future projects will have an “emphasis on an interdisciplinary approach, acknowledging how difficult it often is to keep an overview across the disciplines.” This is intriguing given the importance of interdisciplinarity and the challenges of tracking concepts in new, unfamiliar fields.
Reviewers of the book argue that it’s not actually a book because it lacks a narrative, a integrating storyline. Agreed. But frankly our definition of “a book” has always been, and remains, fairly elastic. So, it’s a book; just a different book. And it’s a very interesting book at that.
Algorithmic decision-making arising from machine learning is ubiquitous, powerful, often opaque, sometimes invisible, and (most importantly) consequential in our everyday lives.
Machine learning (ML) is critically important for libraries because it offers new tools for knowledge organization and knowledge discovery. It also, however, presents significant challenges with respect to fairness, accountability, and transparency.
I believe that artificial intelligence will become a major human rights issue in the twenty-first century.
Safiya Noble (2018). Algorithms of Oppression.
This blog will attempt to chart ML developments and issues in libraries and to identify trends in the wider AI community that impact libraries.
“The danger is not so much in delegating cognitive tasks, but in distancing ourselves from – or in not knowing about – the nature and precise mechanisms of that delegation”
de Mul & van den Berg (2011). Remote control: Human autonomy in the age of computer-mediated agency.
Libraries have often been instrumental in championing new technologies and making them more accessible. As we adopt and develop ML tools and services, something I think is an imperative if we are to advance our mission, we also need to be aware of the emerging “new digital divide”:
A class of people who can use algorithms and a class used by algorithms.
David Lankes (Director, SLIS, Univ. of Southern Carolina).
Looking forward to this journey. Let me know what you think.