Introduction
If we regard schools, colleges and universities as institutions that process information, the management of data represents the first step in many that enables these institutions to deliver education services to their local and wider communities. They start by distilling data into information, information into knowledge, knowledge into wisdom, and wisdom into actions. However, as the volume of data rises it becomes increasingly difficult for teachers, student support teams and admin teams to convert data into information, knowledge, wisdom and actions which enable them to support the myriad of students in their respective institutions. In this short article I would like to detail the use of oracles and machine learning agents which could help schools, colleges and universities to capitalise on learning analytics to enhance teaching, learning and assessment; as well as other student facing services.

Oracles
At the present moment in time Ada, Bolton College's Digital Assistant and other complimentary information systems and business intelligence layers across the College function as oracles. Oracles respond to questions or queries that are posed by students, teachers and support teams across the College. A number of the advanced oracles at the College do not wait for a student or teacher to pose a question; rather the information is presenetd to students and colleagues via the Student and Staff Home Pages. You are already familiar with these types of oracles which manifest themselves in the form of push notifications on your smartphone. Students at Bolton College receive regular push notifications which notify them about a whole manner of topics such as:

• events that are happening around the College;
• exam notifications which include exam titles, dates, times and room numbers;
• deadline dates for assignments and coursework;
• work placement information which includes work placement locations, dates, times etc;
• deadline dates for agreed actions and targets; and many more.

The use of oracles in the education sector are representative of a great set of services which provide valuable information and actionable insights to students, teachers and support teams. For example, Ada - Bolton College's digital assistant is able to respond in a contextualised and personalised fashion. We are looking forward to observing Ada and how she will interact with everyone on the campus; especially as she improves over coming months and years; and as she manifests herself as a true virtual digital assistant to all our students, teachers and support teams. Enabling Ada to have a degree of agency will be the next key milestone.

How will machine learning agents support the student?
I thought it would be useful to detail three scenarios were machine learning agents could support students, teachers and support teams. At first glance they may sound rather far fetched but the technology to support the introduction of machine learning agents is already available to the education sector. By combining available technologies such as machine learning, learning analytics, natural language processing, natural language generation, adaptive learning and learning management systems; schools, colleges and universities can transform many of the services that they offer to their students.

Scenario 1:
In the following example Ada answers a question relating to the hand-in-date for an assignment.

Student: When is my next assignment due in?
Ada: Hi Aftab, your next assignment entitled 'Types of Business Ownership' is due in the 3rd of May at 5pm. You will need to secure a distinction grade for this assignment otherwise you will fall short of the points that you need to get into university. Mr Dean Baggaley is running a tutorial on the 'Types of Business Ownership' on the 19th of April at 11am in Room A2.36 at the Deane Road Campus. I hope you find this helpful. Good luck with your work.

The response that is given by Ada is a complex one because she is able to respond to the student in a contextualised fashion. Ada can only do this if she works in conjunction with other machine learning agents that are employed by the College to support each student. In this example, the response that is given by Ada is informed by the Assessment Agent (which monitors assignment deadlines, student grades, test scores and more), the Learning Support Agent (which monitors student support needs) and the Careers Agent (which monitors and supports student progression paths).

Scenario 2:
In the following example Ada initiates the dialogue with the student without any prompt from the student.

Ada: Good morning Aftab. I hope you enjoyed all the sunshine that we had this weekend. I see that you have an assignment on 'Motte and Bailey Castles' that is due in on the 2nd of May. You have secured good merit grades in your last two assignments and I think you are well placed to secure a distinction grade for the Motte and Bailey Castles assignment. Would you like to aim for a distinction grade?
Student: Yes, why not.
Ada: That's great. The following online tutorial on Motte and Bailey Castles is a popular one. I have also reserved a couple of books for you in the library. If you need further help with the topic just ask me.

In this dialogue, Ada has the support of the Instructional Design Agent (which compiles personalised and contextualised online tutorials), the Assessment Agent and the Library Agent (which facilitates access to library services). Ada even comments on the local weather.

Scenario 3:
If students consent to digital assistants monitoring their correspondence with their teachers we could see the rise of other support services.

Ada: Hi Aftab. I noticed that in your recent messages to your teacher you thought that you were worried that you may have chosen the wrong course. If you need advice and support please contact the student support team in Student Services. It's not too late to change courses.
Student: Can I do a Level 3 Art and Design course?
Ada: There are 2 places left on this course. You meet the entry qualifications for the course. The following link provides more detail about the course. Would you like to talk to someone?
Student: Yes please.
Ada: A member of the Student Services team is free this afternoon at 2pm, 3pm and 4pm. Please select a time that suits you.

In this dialogue Ada identifies a support need and offers options to the student. The service takes advantage of various datasets such as course listings, vacant places on a course and the Student Service calendar.

In all three of these scenarios Ada and the machine learning agents that support her have moved well beyond the traditional oracle. Ada behaves as a semi-autonomous agent that prompts, guides, directs and supports each student through their studies. These machine learning agents never sleep, they are always available to support the student, they constantly monitor every student who is enrolled at the College, they learn from their mistakes, they constantly improve and they behave cooperatively with their fellow agents. The advent of natural language generation means that Ada and the other machine learning agents will be able to provide regular written reports to teachers, teaching assistants, support teams and to parents about the progress of each and every student on the campus.

The ILT Team at Bolton College has been encouraged by the preliminary research that it has undertaken on natural language generation. We could soon be in a position were teachers will never have to write a student report again because they will be produced by a team of machine learning agents - reports that are a free of spelling and grammatical mistakes, free of factual errors or missing information. Reports that are always produced on time and at scale.

Summary
In time, machine learning agents or virtual digital assistants will manifest themselves everywhere in an education setting. They will be developed by vendors to support the use of their management information systems, learning management systems, library management systems, building management systems, cashless catering systems; concierge services, school, college and university websites; by educational publishers, careers support services and more. Getting proprietary machine learning agents to cooperate with one another will be a major challenge for the future.

The advent of the digital assistant in the education sector will provide schools, colleges and universities with the opportunity to enhance many of their student facing services. Larger institutions will be able to provide improved services at scale - something that was not possible prior to the advent of the digital assistant.

One of the most compelling aspects of digital assistants manifests itself when they are able to support a pupil in a school, a student at a college or a student at a university throughout their entire stay at the institution. In the ideal scenario a digital assistant could support a child from pre-school, junior school, high school, through to university and beyond. However, the fragmented nature of the education landscape will mean that the student will have to engage with multiple agents or digital assistants throughout his or her studies.

As I mentioned earlier the digital assistant or agent is always seeking to improve. Each interaction with the agent, each click on the student home page and each click on the institution's learning management system provides it with additional knowledge to support current and future students at the school, college or university. The agent will capture every online tutorial that you engaged with, it will capture data on every video that you watched, it will record all your test results, it will monitor your learning preferences during the course of your studies and much much more. As the acquisition of this data turns into knowledge and wisdom; schools, colleges and universities will be better placed to support the needs of each student.