INTRODUCTION
The early twentieth-century librarian Shiyali Ramamrita Ranganathan's fifth law of library science – ‘a library is a growing organism’ – highlights the mutable and organic nature of libraries and librarianship. In the section titled ‘Future’, he muses on developments beyond the scope of his own life and career:
What further stages of evolution are in store for this growing organism – the library – we cannot anticipate fully. Who knows that day may not come when the dissemination of knowledge, which is the vital functions of libraries, will be realized by means other than those of the printed book.Footnote 1
Similar to Ranganathan's musings in 1931, today, over ninety years since the publication of his five laws, we, too, cannot fully understand the impacts that new technology – underpinned by Artificial Intelligence – will have on both the profession and on the professionals working within it. One way of attempting to understand it is to explore what Artificial Intelligence actually is.
DEFINITIONS AND CONTEXT
Artificial Intelligence or, rather, the term Artificial Intelligence (hereafter referred to as AI) has no place in serious business. So began Robin Chesterman's presentation at the 50th annual conference of the British and Irish Association of Law Librarians (BIALL) in June 2019. He outlined that the term AI muddies society's understanding of the technology which it encompasses, as well as its use and application.
The term Artificial Intelligence was coined in 1955 by John McCarthy and his team in their proposal for the 1956 Dartmouth Conference. Their aim was to find out how to make machines use language, form abstractions and concepts, and solve problems typically reserved for humans – in short, the simulation of human intelligence via technology. To an extent, AI, in the mid-twentieth century, was a framework: it was a theoretical idea and a set of objectives set-out to be achieved and accomplished.
Returning to Chesterman's remarks made in 2019 about the confusion around the meaning of AI, McCarthy's framework is useful today by way of helping us understand what AI is: not some unexplainable, blanket, futuristic technology that will bring about the death of librarianship, or other longstanding professions; but rather, a puzzle to be solved through the application of existing and newly-emerging technologies.
Today AI, in its simplest form, tends to be divided into three main areas:
1. Machine Learning (ML) – where technology, underpinned by algorithms, detects patterns in data, and applies these new patterns to automate certain tasks.
2. Natural Language Processing (NLP) – the branch of AI concerned with giving computers the ability to understand text and spoken words in much the same way human beings can.
3. Knowledge Reasoning (KR).
The UK government's recently-established Office for Artificial Intelligence, in its September 2021 report, National AI Strategy, acknowledges that:
‘Artificial Intelligence’ as a term can mean a lot of things, and the government recognises that no single definition is going to be suitable for every scenario. In general, the following definition is sufficient for our purposes: ‘Machines that perform tasks normally requiring human intelligence, especially when the machines learn from data how to do those tasks.' Footnote 2
As outlined above, most definitions of the term AI tend to be broad. Luke Tredinnick posits that ‘most current [AI] practice is focused on addressing specific problems in specific contexts rather than in simulating general human intelligence.’Footnote 3 Tredinnick's comments highlight the difficulties that arise when attempting to blanket-define AI. Rather, professionals, the media, and society at large should invest time and attention to focusing on how it is being used and applied to roles and workloads in specific sectors. The legal sector is among the many that are capitalising and harnessing the power of newly-emerging technologies.
THE APPLICATION OF AI IN THE LEGAL SECTOR
In his first book, Tomorrow's Lawyers, Richard Susskind outlined that the legal world will change ‘more radically over the next two decades than over the last two centuries.’Footnote 4 Despite the enduring features of the legal profession, emerging AI-based technologies are beginning to change the way certain duties are performed. The following examples are, typically, underpinned by Machine Learning and Knowledge Reasoning:
• Document analysis (to streamline the transaction processed in various sectors such a mergers and acquisitions and transfers of estates)
• Contract intelligence (to enable a whole suite of documents to be scanned into an AI system to spot areas of high and low risk, aid the wording process, and, more recently, ensure GDPR compliance)
• Clinical negligence analysis (examines the outcome of previous cases to predict the outcome of a claim)
• Case outcome prediction (which tends to be the preserve of legal practice in the United States)
While these technologies are having a profound impact on the way solicitors in the UK work, it is important to note that AI, in the form of ML, NLP and KR (the latter not as widely used in the UK legal information sector), is only one component of the bigger change that Susskind writes about. It is also amid this shift that legal information professionals will need to consider their own roles in this uncertain future.
Many UK law firms situate their use of technology in the wider context of ‘innovation’ (innovation being one of many interchangeable terms used). ‘Legal innovation is all about embracing a culture of change, cultivating engaged employees, using technology and creating loyal clients’Footnote 5. Technology is not the nexus of innovation; rather, a component alongside people, in the form of internal employees and clients, within a wider context of change.
Figure 1 illustrates AI within the wider scope of legal technology, which encompasses other, non-AI-based technologies such as IT and finance systems; library and knowledge management systems; risk and compliance; human resources; and casework flow software. To assume that AI is the sole component of innovation, and that all legal technology is AI (AI and legal technology are not interchangeable terms), is a fallacy. Yet, keeping Branum's remarks in mind on the intangible aspects of innovation – the knowledge and skills employees within the firm hold – what role do information professionals play?
Figure 1 Graphic of what innovation looks like in a commercial law firm.
THE ROLE OF LEGAL INFORMATION PROFESSIONALS
In an article for the American Association of Law Libraries’ journal, Spectrum, Sherry Xin Chen and Mary Ann Neary posit two statements:
Law librarians, as information professionals with a unique understanding of users' search habits, goals, and available data, can help the institution tailor the application and maximize the benefits on the AI system
The time is ripe for law librarians to incorporate background knowledge of both database algorithms and AI corpus contents into the legal research curriculum in both academic and firm instructional settings Footnote 6
While both statements encouragingly advocate for the continued presence of information professionals in the legal sector, neither present any specific, practical examples as to how information professionals are contributing to the implementation and use of AI in their places of work. This is problematic because not only does it create further confusion among information professionals on how they can contribute, rather than can they contribute; it reinforces the AI hype discussed earlier.
Neary and Chen, instead, focus their study specifically on the role of Natural Language Processing, and where legal information professionals’ skills can be best implemented: ‘librarians, aware of what results can optimally be retrieved by a particular search, can gauge the gaps or weaknesses in an AI system by evaluating results’.Footnote 7 Yet, no case studies are offered in support of this statement.
A UK-wide example, where NLP is often used in legal research, is the use of databases, such as Westlaw UK. This particular database enables researchers to look for information using a range of search terms and connectors, as well as the option to input questions using natural language in plain English. Further developments to NLP include Practical Law Dynamic Tool Set, which includes searching underpinned by advanced NLP, giving the user a more Google-like search experience, using natural language rather than constructed searches using terms and connectors.
In their chapter on AI and its place in the corporate library, Ed Walters and Sean Tate outline: ‘if information professionals themselves use AI tools to conduct their own experiments in supervised learning, and learn how to gain new insights from legal data, there is a great potential for a broad-based renaissance in legal services, in which everyone has the opportunity to participate’Footnote 8. While reassuring and, to an extent, hopeful, such participation is dependent on a number of important factors:
• The law firm has the economic infrastructure for such initiatives;
• There is complete and equal cohesion between all key departments;
• Time is set aside that will enable key professionals to prepare, plan and execute the initiatives, and be afforded the time to reflect upon them.
While there are important innovation-based initiatives taking place in law firms involving the cooperation of information professionals, the extent to which information professionals are involved in such projects is still largely unknown; or, as is often the case, minimal. This, in part, is resultant from commercial secrecy – law firms are, on the whole, reserved about disclosing exactly which technologies they are using because of competition among firms, and, in some cases, protecting the identity of their clients.
TRANSATLANTIC ANALYSIS
In the 2019/20 BIALL Annual Law Firm Survey, which gathered 56 responses from law firms across the UK, ‘law firms libraries selected, on average, over five issues being face in the current financial year’Footnote 9 – one of which was the ‘use of Artificial Intelligence’ raised by 25% of respondents.Footnote 10 No additional information was given as to what ‘use’ means and what the ‘issue being faced’ refer to: the threat of job loss? Gaps in information professionals’ knowledge and understanding of AI systems? General uncertainty across the profession regarding the future impact of newly-emerging technologies? It would be interesting to see further work carried out in this area, where UK-based information professionals are concerned.
Across the Atlantic, the American Association of Law Libraries (AALL) published their State of the Profession Report, which included responses from 229 US-based corporate legal departments and law firm libraries. The report summarised the following:
• 28% of law firm and in-house information professionals are planning or creating one or more AI/ML initiatives;
• 52% said that they didn't have an AI/ML initiative and had no plans to start one;
• 11% said AI/ML impacted their library's workflow;
• 43% plan to acquire or further develop their AI/ML skills in the next two years.
While 52% of information professionals outlined that their workplace does not have any AI/ML initiative(s), or any plans to start one, a minimum of 43% of information professionals have plans to upskill and develop their AI/ML skills in the next two years. This highlights an awareness of the importance of AI/ML skills among a large proportion of the profession, and is something that US information professionals are not only taking note of, but that they are also beginning to action.
While the picture of US legal information professionals’ engagement with AI is clearer than it is in the UK, the Chartered Institute of Library and Information Professionals (CILIP) published, in May 2021, The Impact of AI, Machine Learning, Automation and Robotics on the Information Profession. The report seeks to help CILIP and its professional community understand how AI, machine learning, process automation and robotics are either already impacting the daily work of information professionals or likely to do so in the near future. This report, which will be looked at further in due course, is both needed and timely.
OPPORTUNITIES
The vast majority of UK legal information professionals are aware of the two main databases used regularly in legal research: Lexis®Library and Westlaw UK. The latest BIALL Law Firm Library Survey outlined that 94% of participants subscribed to both of these databases. Information professionals’ engagement with them range from using them in research on behalf of legal professionals; to teaching / instructing them how to navigate and use them effectively.
In relation to the latter point, I would like to draw on the work of Dominique Garingan and Alison Jane Pickard whose recently-published article explores the theoretical frameworks for algorithmic literacy in the legal information profession. Their timely research draws on a plethora of studies and articles, one of which posits that ‘the danger of reliance on artificially intelligent systems is not so much in the increased delegation of cognitive tasks to these systems, but in information professionals and information users distancing themselves from, and not knowing about, the nature, precise mechanisms, and repercussions of that delegation.’Footnote 11 As evidenced in the BIALL Survey, this ‘distancing’, in part, feeds in to the ‘issues being faced’ by information professionals regarding the technological aspects of their roles.
In direct response to these issues, Garingan and Pickard go on to outline that: ‘just as librarians’ roles may involve evaluating technologies and providing feedback as part of procurement processes, the traditional skills that underpin librarians’ professional training will require pivoting to help develop new algorithmic literacy initiatives.’Footnote 12 As highlighted by Neary and Chen, the time is now ripe for information professionals to realign their skillsets in order to maximise their full potential in what is now being referred to as the fourth industrial revolution.
As more law firms invest in AI-based technologies underpinned by ML, NLP and KR, there will be a greater anxiety around the use of these technologies from an ethical and legal standpoint. There is a growing demand for AI technologies to be far more explainable as AI rapidly becomes ubiquitous across many sectors. In their 2018 paper, Amina Adadi and Mohammed Berrada outline that ‘Artificial Intelligence (AI) is democratized in our everyday lives’,Footnote 13 yet, the majority of society – be it in the UK or the US – is no closer to understanding how AI works, and, more importantly, what level of understanding those using it should have with regards to how it achieves the results produced.
There is no one single definition of explainable AI (XAI); however, in the context of, for example, contract intelligence (explored above), ‘the goal of enabling explainability in ML [Machine Learning] is to ensure that algorithmic decisions as well as any data driving these decisions can be explained to end-users and other stakeholders in non-technical terms’.Footnote 14 Yet opacity – the lack of ‘any concrete sense of how or why a particular [answer/result] has been arrived at from inputs [at its most basic]’Footnote 15 – ‘seems to be at the heart of new concerns about ‘algorithms’ among legal scholars and social scientists’.Footnote 16 For those using AI-based technologies, the legal professionals or other stakeholders, i.e. the end-user, there are still concerns regarding the opacity surrounding the explainability and understanding of how certain results are achieved and, more importantly, if those results are correct.
The role of information professionals as educators is no new phenomenon. Traditionally in UK law firms, information professionals have played an instrumental role in the provision of legal research and database training to legal professionals for decades. In the 2019/2020 BIALL Annual Law Firm Survey, all respondents outlined that their law firm offered training in some form or another. 100% of respondents who said they provided training, bar one respondent, outlined that research training was facilitated by library staff, and 71% by vendors.
I would argue that where academic information professionals have a duty to facilitate the teaching of information literacy in all its various forms, the legal information professional now has a duty to implement and facilitate training on the use AI-based systems for legal staff. In light of Burrell's and Carabantes’ work, as well as the statement posited by Neary and Chen, information professionals’ role of training staff on how to use these technologies brings to the fore two important issues that need addressing:
• Where do information professionals sit on the opacity scale?
• In light of the latter, individuals would, in some cases, need to not merely up-skill; but, potentially, re-train in a completely new discipline.
To better understand the first problem, I propose the following diagram to explain where on the opacity scale three core individuals sit: Figure 2
1) The technologist (the creator of the AI);
2) The implementer (legal information professional);
3) The end-user (legal professional).
Figure 2 Graphic showing where three sets of professional sit on the opacity scale.
There is a noticeable decrease in opacity the farther up the scale the person is situated. The AI expert does not reach the top of the scale because, as analysed by Burrell, there exists opacity in the form of cognitive mismatch: possibly the most concerning type of opacity, which ‘prevents the engineers who develop certain ML [Machine Learning] models to understand how their own creations work’.Footnote 17 It is their responsibility to impart their knowledge of the AI to the implementer, the person/team responsible for choosing the AI and implementing it in their organization, and overseeing any issues that arise.
In turn, it is the role of the implementer to train the end-user in how to effectively use it for their needs. The implementer's knowledge, on the opacity scale, sits somewhere between AI expert and non-AI expert: their understanding is enough to train others to use it, but not, necessarily, for them to create it. In the majority of circumstances, there will always be a gap in knowledge between the information professional and AI expert. As highlighted in the statistics provided by the AALL survey, there will often be a collaboration between information professionals and information technology professionals, in which case, institutional opacity will fluctuate, and individual opacity will be entirely dependent on the training and support professionals receive.
As illustrated in the above diagram, and Carabantes’ analysis, the legal professional will understand that, for example, when 1,000 documents owned by a specific entity are uploaded to a contract analysis system, the software will scan each of those documents and outline what it is and what it contains through a long process of Machine Learning. The legal professional understands the output and what the end result might, or should be, but might not understand how this is achieved. This is where information professionals are critical.
As mentioned, the symbiosis between roles and departments – legal information professional, legal professional; information technology professional, etc – requires reconfiguring within organisations as a whole. This is being demonstrated by one of the UK's leading City-based law firms, Linklaters LLP. In an interview in Information Professional magazine, CJ Anderson, former Head of Information and Research, outlined that:
Information professionals operate as account managers with skills and confidence that was not necessary a few years ago […] they are also free to work with people in other departments. At the moment, there is someone in my team who owns a couple of our internal databases who has spoken to our data architect – in our technology architecture team who is responsible for how all the databases join up. Footnote 18
Linklaters is one of a small number of reported cases where information professionals are not only cross-collaborating with other departments within the organisation; but are taking proactive measures in embedding their roles in the technological processes at play. However, in order to achieve this on a large scale across the UK, technological illiteracy must, firstly, be recognised as a skills gap within the majority of corporate law firms – as well as within other institutions across all sectors – and, secondly, must be remediated through various channels.
Carabantes outlines that ‘technological illiteracy is remediable, either through a new honest journalism ethic committed to the truth over the interest of the advertisers, or thanks to a new revolutionary school capable to penetrate to the depths of the limbic system’.Footnote 19 This, I would argue, extends to the role of information professionals within law firms, who have the capabilities of understanding users’ searching habits; the legal research process, and in-depth understanding of legal databases, and, therefore, are able to assist legal professionals in the use and understanding of new legal technologies.
CONCLUSION
Mere suggestion that a fundamental shift within the information profession needs to happen is not enough. Garingan and Pickard highlight that ‘the importance of introducing legal technologies into the law school curriculum’. There is an argument to be made for this introduction of legal technologies, and AI, to be extended to library school curricular across all iSchools in the UK and US to better-prepare newly qualified information professionals. This will enable them to feel confident in evaluating, using, and teaching others on how to use AI-based technologies they might encounter in their future careers.
For mid-career, experienced, and management-level professionals, upskilling and refreshing current skillsets is an important component of continuing professional development. To ensure that information professionals, especially those engaged in the Chartership process with CILIP, recognise gaps in their knowledge of technology, changes must be made by professional bodies supporting those individuals and organisations. CILIP's Professional Knowledge and Skills Base (PKSB) outlines the various literacies information professionals across many sectors work with, one of which is digital literacy, which could extend to algorithmic literacy. Andrew Cox's report highlights:
One important implication for all information professionals is that a basic form of literacy that they need is understanding of AI and robots: they need AI and robotics literacy. Algorithmic literacy could be seen as a subset of this […] AI, robotics and data literacy is something that libraries can play a role in promoting. Footnote 20
Such a move would also need to be adopted and propagated by other organisations, such as BIALL, and its international counterparts. The 10-year-old BIALL Legal Information Literacy Statement could accommodate such changes within Research Skill Five: Continuing Professional Development – refreshing the legal research skills required of a modern lawyer.
While there is still work to be done, information professionals ‘are akin to the inhabitants of a small isolated island who have just invented the first boat, and are about to set sail without a map or even a destination. […] The inhabitants of our imaginary island at least know that they occupy just a small space within a large and mysterious sea.’Footnote 21 As a profession, we are aware of the challenges ahead, but not, specifically, their full form or impact. In order to preserve Ranganathan's fifth law regarding the organic nature of librarianship and its professionals, we must continue to capitalise on the opportunities that present themselves.
