Recent Events

Author Archive | Don Hawkins

Onboarding AI and Machine Learning

Brian Pichman

Brian Pichman

Brian Pichman, Director of the Evolve Project, said that AI is the theory and development of computer systems able to perform tasks that normally require human intelligence; it is becoming more available to the masses. Machine learning is a subset of AI that uses algorithms and statistical models to allow a computer system to make decisions around a task without explicit instructions, relying on patterns and inference.

AI has 4 components: it is logic-based, rules-based, pattern-based, and includes deep learning and neural networks. It exists with self-driving cars, etc. (pattern-based); Strong AI does not exist yet.  Pattern recognition means to create an algorithm able to teach itself without any external help. Voice recognition systems can respond and remember a user based solely on their voice.  Samsung has released bots to monitor air quality, place orders, etc. We use AI for optical character recognition, to advance user preferences (recommendations), and sensory data analysis.

AI is used in healthcare to identify and notice predictable trends, read X-rays, and gather information. Here is a good article on possible manipulations.

Other applications:

  • Smart homes can issue an alert to distinguish a person vs. an animal, know that it will rain tomorrow so there is no need to water the lawn today. Robots are coming out for home use.
  • AI has the potential to replace many minimum wage jobs, especially in the service industries. There is already a burger-flipping bot.
  • ROSS can make legal research faster and easier.

Challenges include precision, context, and training. It is important to make sure that the data provided is correct and useful.

AI in libraries:  There are many skills to learn. AI can do simple tasks so you can focus on other things. There is some impact on staff. How do we handle simple reference questions? We can make jobs more meaningful and get rid of tedious work.

AI does not have emotional intelligence and cannot show empathy; decisions are based on facts.


Bot Literacy: Teaching Librarians to Make Twitter Bots

Mark Eaton and Robin Camille Davis

Mark Eaton and Robin Camille Davis

Mark Eaton from Kingsborough Community College, CUNY, and Robin Camille Davis, John Jay College of Criminal Justice, CUNY, described a workshop they developed to teach librarians how to make Twitter bots. They began with first principles:

  • A bot is a little program that does the same thing over and over again, with a pre-programmed variation.
  • A Twitter bot usually sends out a thematic tweet on a regular basis. The tweet is created using a script and some data.
  • Manipulative bots don’t identify themselves and retweet a political slogan over and over. 

The workshop emphasized becoming familiar with the Python programming language. It was presented three times at conferences. There are many claims that coding is a new literacy; today the definition has expanded from traditional notions of reading and writing to the ability to learn, comprehend,and interact with technology in a meaningful way. In their workshop, they wanted participants to learn how to build bots, what tools are needed to learn to make bots, how to deploy a bot, and the social implications of bot making.

The Python computer language has a low barrier to entry, so it is good to start with it. Writing code is all about problem solving. Don’t learn to code; code to learn. Consider the social implications of what your bot does. Aim for familiarity, not fluency, at first, and make it easy to begin running code, then make something fun that users can show off to other people. Provide a glossary and write your own plain-language definitions.


  • Bot literacy helps us to better understand our networked world.
  • Making bots is an effective entry point into programming.
  • A scaffolded approach is a god way to teach programming.
  • Include creative time.

The workshop was successful; participants provided favorable feedback.


Open Access (OA): Latest in the Landscape

Katharine Dunn

Katharine Dunn

OA began by providing readers with free access to journal articles, but it has changed significantly recently. Katharine Dunn, Scholarly Communications Librarian at the MIT Libraries, reviewed recent developments, which include OA agreements (and disagreements) with publishers. She said that Europe is a hotbed of activity because they are more willing to coordinate their resources and have thrown the weight of their funding behind gold OA.

A major development is the cancellation of subscriptions to publisher’s databases, notably Elsevier’s, because their demands are acceptable to the community. There is a lot of momentum in pushing the publishers and walking away. Users want immediate open access to articles, fair article processing charges, sustainable costs, and full transition to OA.

Recently, the University of California terminated their agreement with Elsevier and issued a document written by the faculty (not the library) explaining their position and why they terminated their agreement. (They were paying $11.5M/year and Elsevier wanted to raise the price.) However, they committed to continuing negotiations, so access to Elsevier’s journals will continue in the interim. The faculty also issued a “Declaration of Rights and Principles Governing Scholarly Communication“.  The UC Berkeley library also published some suggestions for gaining alternative access to Elsevier articles.

Alternative Access to Elsevier Articles

Other developments:

  • Recently, Temple University also terminated their agreement with Elsevier.
  • The Royal Society of Chemistry (RSC) and MIT recently signed the first North  American “Read and Publish” agreement to shift publishing models toward OA.
  • Plan S is a group of national level funders in Europe wanting full and immediate OA.  OA is foundational to the scientific enterprise; a decisive step towards the realization of full OA needs to be taken now. Plan S is not yet fully formed. It is not clear what a compliant platform is.

Academy-owned publishing: We need to think about ways that academies can support and sustain their publishing activities. Many of them have hired a corporation to provide services, but the corporation’s interests are not aligned with those of scholarship. We need to establish good practices for vendor negotiation and keep technology and journals community controlled.

Academic led publishing day, an inaugural event t0 celebrate and facilitate academic-led publishing was recently held on February 7, 2019.

Going forward, MIT formed a task force to strengthen and increase OA at MIT.  Some of its recommendations include:

  • Adopt an all-campus OA policy and an OA policy for monographs.
  • Consider adding an OA requirement to all existing and new internal grants.
  • Create an Open Textbook Fund and an OA Infrastructure fund.
  • Develop a set of open licenses for software.
  • Double down on responsible ways to manifest MIT’s foundational belief in the value of open sharing.

Website Design Winners and Losers

(L-R) Jeff Wisnewski, David Lee King, Marshall Breeding

(L-R) Jeff Wisnewski, David Lee King, Marshall Breeding

Speakers in this session examined some library websites and commented on them. David Lee King from the Topeka & Shawnee County Public Library in Kansas began with the Seattle Public Library which has recently modernized its website. He said that in general they have done a good job. You can find all the major functions and services of the library, and it is mobile friendly. They developed the mobile site first but don’t seem to have considered other devices as much. They have a big empty box at the left with a design in it, which is using up a lot of space.

On the Topeka & Shawnee County Library’ site, news is shown as a big blog in the middle of the page. Hours are in the footer. People are finding the databases and services.  The site search engine will be changed. They had been using Google’s search appliance, but Google has discontinued it and now defaults search to the main Google site, so sometimes they sometimes see Google ads on their page. To avoid this, they will be changing to WordPress.

Marshall Breeding, Independent Consultant, Library Technology Guides, concentrated on the technologies which may run counter to library values and strategies which he has observed from his years of maintaining Key issues are security/privacy and branding. Browsers using http instead of https are flagged as not trustworthy because data transmitted using http are not encrypted and can  . Any browser request to http should be redirected to https. 56% of library websites are not secure, 48% of public library sites. Some library websites connect to ad trackers. Use tools to understand what is inside your pages.

General observations:

  • Library websites vary from exceptional to mediocre. Technical underpinnings are often weak. Some are not mobile friendly.
  • Larger libraries have more budget for staff to fix problems.
  • The library is the brand. Brands are not highlighted as they should be.
  • URL branding: It is important to have a distinctive brand for the library. Expected domains are .org, .gov, and .us.
  • Other URL issues: keep it simple, avoid URLs on root pages.
  • Should a library build a website itself or use a managed platform? Here are some considerations:

Jeff Wisnewski, Web Services Librarian, University of Pittsburgh, took the Harvard University Library as his test site after his Director recommended it as a “nice site” to him.


  • The homepage is very visual. It has a clear call to action; programs and services are identifiable; it is up to date and has meaningful graphics. Search options are clear and understandable. But the search box searches the library’s website but not its content. Users wanting to search content are directed to Hollis (the catalog).
  • Navigation links and labels contain “trigger words” (have “scent”) that users will look for to achieve their goal. Choices are ordered in the most logical or task-oriented manner.
  • The content is jargon-free. Pages make good use of headings, are error free, and use plain language.
  • Mobile usability: Is it usable on mobile, responsibly designed, quick loading?
  • Trust and credibility: Is there a real organization behind the site (i.e. a physical address or photo of the office)? Are interactions available? Is the content fresh and frequently updated?