AI Literacy for Lawyers - Key Concepts & Terms Explained

Introduction

You do not need to understand how a combustion engine works to drive a car, but you do need to know where the petrol goes and that running it dry will stop you in your tracks. Generative AI is already used in most legal practices, yet many lawyers are effectively ‘driving blind’ when they type into ChatGPT or Claude. That gap matters because it explains why the same system can produce excellent work one moment and confident errors the next.

A large language model does not reason like a human; it predicts the next word from patterns in data, a ‘clever parrot’ rather than an understanding mind. Once you see that, its behaviour becomes predictable. Tokens and context windows determine how much information it can process at once. Knowledge cutoffs explain why outputs can be out of date. Web tools help but do not fully solve this. Hallucinations are structural: the system cannot tell when it is wrong. Temperature controls variability, meaning identical prompts can produce different answers. Prompting is therefore a skillset, including system prompts, chain-of-thought, few-shot examples and prompt libraries. Above all is verification, the key professional safeguard.

This short webinar turns those concepts into practical understanding, not abstract theory. Each is framed as a predictor of real behaviour in legal work: what to expect, when to trust outputs and where failure occurs. You will leave with a clear mental model, a usable glossary and three practical techniques to improve accuracy, control outputs and reduce risk.

Professional competence now includes competent use of AI tools. The Solicitors Regulation Authority Code of Conduct applies in full, and the High Court in Ayinde v Haringey has confirmed that AI use requires supervision equivalent to a trainee. Understanding the system is therefore essential to using it safely, effectively and in line with professional standards.

What You Will Learn

The webinar will cover the following:

• What a large language model is and does - the ‘clever parrot’ explained in plain English
• Tokens and the context window and what changes when documents get too long
• Knowledge cutoff - what the model cannot know, what web search improves and what it still cannot fix
• Hallucination - what it is, why it happens and why the model cannot reliably flag uncertainty
• Temperature and the practical habit of running prompts multiple times to compare outputs
• Prompts and system prompts - what is the difference and why system prompts are the most powerful control layer
• Chain-of-thought and few-shot prompting - two techniques that reliably improve legal output quality
• Prompt libraries and tool selection - general vs legal-specific tools and why verification is a core professional duty

This pre-recorded webinar will be available to view from Thursday 14th January 2027

Alternatively, you can gain access to this webinar and 2,400+ others via the MBL Webinar Subscription. Please email webinarsubscription@mblseminars.com for more details.