Thought I'd re-post the introduction to AI and the legal profession here. We hope to do a guide in the future, but in the meantime, here are Jaysen's draft notes on a live webinar hosted by the European Legal Tech Association (ELTA)).
And thanks again to Jonty - I understand that you spotted the webinar in the first place!
Draft notes:
Introduction
And thanks again to Jonty - I understand that you spotted the webinar in the first place!
AI and the Legal Profession
Host: María Jesús González-Espejo, managing partner of EmprendeLaw and founding member of the Instituto de Innovación Legal
Speakers: Richard Tromans, founder of Artificial Lawyer and Dr Francisco Martin, co-founder and CEO of BigMLHost: María Jesús González-Espejo, managing partner of EmprendeLaw and founding member of the Instituto de Innovación Legal
Draft notes:
Introduction
- People underestimate machine learning. That's what we should look into in the short-medium term (machine learning is a form of AI where a program is designed to learn on its own).
- Machine learning can be used to solve complex problems i.e. NDA Lynn: automatically determines whether you should sign NDAs, which sentences are deal-breakers or not acceptable.
- Many tasks performed by lawyers can be automated and it’ll be used to achieve human-level performance at a higher speed for a cheaper price.
- Lawyers in most law firms complete their work manually with the help of IT for a range of tasks like document management, communication, billing and virtual data rooms etc.
- There are some new developments in assisted technology. For example, contract express, but this is more to do with automation.
- The growth of machine learning and natural language processing (allowing a machine to understand natural languages) will have a dramatic impact on lawyers and in-house teams. The whole business relationship will change.
- For example, in M&A due diligence, natural language processing can help lawyers look for specific clauses or evaluate the risks.
- Expert systems like Dealogic (which I’ve used at my law firm) uses inputs from lawyers to form a machine. That alters the way in which lawyers work because the data is available immediately at all hours.
- Some companies are entering the litigation prediction space.
- Automation increases efficiency and productivity, its great for big law firms, and if small law firms can afford it, it will likely level the playing field.
- It’ll allow lawyers to focus on high-level work.
- It’ll affect the cost– typically a client will pay a fixed fee for a due diligence exercise, if you have an automated system such as Slaughter & May’s Luminance which reduces cost - you can increase profit margins.
- AI can also give new value, for example, traditional tasks which weren’t worth doing because the risks didn’t justify the costs could be made possible: this can create new and improved client relations.
- One of the speakers, Richard Tromans, was speaking to Google yesterday. Google performed a test where AI beat lawyers in their document review analysis/predictions. It was interesting to hear how lawyers overestimated their abilities!
- Lawyers/law firms not using AI may be in a riskier position or simply less accurate.
- Law firms need to think about how they’re using data. They should start collecting it and consider how it can be automated. Every time you a lawyer reviews a contract, all that knowledge can be used in machine learning to automate tasks or check clauses.
- Lawyers may need to change their mindset. They could start looking at contracts as data rather than about its textual meaning.
- A 2017 survey of the UK 100 law firms found the top 30 are using an AI/machine learning system – either for fully paid client work or in a pilot.
- Davis Polk in the US is now using Kira to do due diligence. Clients – investment banks – are reluctant to pay associates for this work and want a better solution. When a firm like Davis Polk, one of the best law firms in the world does this, that sends a message to the market.
- This is happening all over the world, Europe has interesting AI companies like French-based Predicted (litigation analysis), German-based Leverton (real estate AI). This extends to Singapore, China and Australia.
- It’s growing really fast. Two years ago, about 5 law firms in the world were using machine learning for due diligence, risk review and compliance, now it’s a huge number (and many aren’t speaking about it!).
- There’s a huge market for automation and a big commercial case, including with small law firms. Soon that cost will go down, and they can specialise in a certain area if that firm, for example, doesn’t offer M&A.
- Smaller firms are also thinking creatively about they can use technology to add value, for example, one firm hard-coded an interactive legal bot. There’s also simpler areas that they can automate such as standard contracts. The higher you go up – that’s where machine learning applies – if you want a like-human capability.
- It's likely that legal innovators – usually 2 or 3 partners at firms will lead this change (as well as external parties pitching). Some firms, such as A&O have already adopted innovation departments.
- Some clients have started to ask law firms about AI systems and how they will benefit the client.
- Not using the right data or using biased data. Tromans mentioned a US case where a company designed an algorithm to work out whether a person should be given parole. It was found to be completely biased against minorities as the designer had misunderstood data. The problem is the court in America started using it – so caution is needed about who are designing these systems and what data is used.
- Subjectivity can be dangerous if applied to litigation suits etc. To overcome these problems, algorithms need to be trained to make fair decisions and a human should still check the final end-product.
- Fully automated law firms will pop up and they will be unbeatable in performance/speed/cost: "the Amazon of the legal services".
- Fully automated back-ends for new companies i.e. corporate matters/financing/NDAs/agreements/terms and conditions.
- All major law firms will adopt AI review systems, litigation systems, for at least 10% of work in the coming years.
- Technology will continue to develop and become more sophisticated.
- It’ll vary based on practice i.e. IP v corporate may use it differently.
- The biggest mover has been M&A and the biggest transactional law firms have now created a new standard: within 18 months every major law firm will be using it – from Davis Polk to Freshfields to Jones Day.
- Litigation prediction tools are also being developed by established legal services like Lexis Nexis.
- Clients will get great value. For example, an insurance law firm can use AI to predict the likelihood of litigation suits, that will help insurance firms to predict their costs and they will significantly benefit. Alternatively, lawyers can offer to review sales contracts/service agreements for clients using natural language processing, and determine what type of customer agreements are most profitable/what will get challenged letter and what the probability of that is.
- Trainees and junior associates might want to consider getting involved in hackathons or groups in cities.
- Many legal AI companies, such as Neota Logic, are currently working with universities around the world.
- Paralegals are likely to get hit first, junior associates may be just enough to get by provided they have enough expertise. There will be a market for legal entrepreneurs.
- Law students at Columbia Law School have already been working on machine learning training.
- Some equity partners may think broadly about how many areas to automate, it may get to the point where – through expert systems and natural language generation – all documents will be automated, and only partners will look at the final documentation: this could be in the next ten years!
