Legal Services Automation & Responsible AI for Lawyers

Welcome to Legal Services Automation & Responsible AI for Lawyers. Follow Programme Reader 02 — Core Concepts & Vocabulary, then practise in the learning system. Sign in and use the study header to open your authorised course. Select this published lesson. Start Course Tutor with Auto audio on: it teaches continuously, answers a typed question, then resumes. Deployment and data location. Software may run on the practice's own equipment, in a hosted environment under contract, or as a public service. A vendor demonstrates a contract review tool. Here are the questions, in the order a practitioner should ask them, with the reason for each. One. Is this a rules engine, an extraction system, a generative system or a combination? The answer determines how it fails and how it must be supervised. Two. Where is the data processed and stored, in which country, and by which entities including any sub-processors? This determines the confidentiality analysis and whether any transfer restriction applies. Three. Is our material used to train your models, and if so can that be switched off in the contract rather than in a setting? A setting can be changed; a contractual term cannot be changed unilaterally. Four. What is your deletion commitment on termination, including backups, and how is it evidenced? Five. For extraction, what is your accuracy on documents of the kind we handle, stated as how often you find what is present and how often you report something that is not, and on what test set? Refusal to answer, or an answer expressed only as a single percentage, is informative. Six. For any generative feature, is the output grounded in a defined document set with references, or is it produced from the model alone? If the latter, it may be used for orientation only. Seven. What audit trail does the system keep, and can we produce it if a client or a court asks how a document was produced? Eight. What happens to our templates and our data if we stop using the service, and in what format do we get them back? Now the internal decisions. Write down which tasks the tool is permitted for and which it is not. Write down who verifies its output and how that is recorded. Decide what the client is told, and whether any engagement term needs amendment. Assess whether the practice is disclosing client material and, if so, whether consent is required or appropriate. Nothing here is advice in any actual matter, and no tool should be assumed to produce any particular saving or result. Ask Course Tutor to explain this step: Four. What is your deletion commitment on termination, including backups, and how is it evidenced? Five. Say what you tried. Pause and try this source task. Why must a data assurance be contractual rather than a setting? The source answer is: Because a setting can be changed by the provider or by an administrator, while a contractual term binds and gives a remedy. Confidentiality and data location commitments that exist only in a configuration page are not commitments to the client at all. Compare your method and result, correct an error, then continue. Use Stop to end Course Tutor.