The Future of Law Post Generative AI (October 2023)
Anyone who tells you they know exactly where Generative AI is taking us is either fibbing or very naïve. This area is changing rapidly and we are seeing developments by the day.
We have discussed where this is going with a few clients and the likely affect Gen AI will possibly have. We thought we would share a few points in this article. This is not set in stone but is the result of some discussions with some very well informed and experienced people across all dimensions of the profession. We have broken this down into six themes:
1. Clients – We see client demand altering. Some work will no longer come to law firms. A number of you will have heard me speak of the example of a law firm partner who shared that his client had a problem but instead of coming to them for advice (which would have taken up to a day), they asked Generative AI to produce something that they looked over and requested that the partner spent 10 minutes reviewing it. We are likely to see more of this, and it will bring a shift to a different sort of work for law firms.
Law firms have always had a mastery of the truth and accuracy and it is foreseeable that they will be asked to “check” a greater number of matters. This will probably also result in work with a higher risk profile. Perhaps more playbook tools will be needed. Anyone reviewing a document from scratch will also know it takes longer than 10 minutes and potentially takes longer than drafting a document from scratch.
On a basic level we are also going to see clients doing more things for themselves. I don’t just mean the above example but creating their own instances of Azure AI and integrating their own documents and precedents. Copilot alone will give them huge capability (especially with resources from the likes of Thomson Reuters being embedded via a plug in).
Also, even where a legal drafting job would legitimately take say one or two days of law firm time, client expectations and what they are willing to pay will reduce. Clients have often thought law firms produce things “at a press of a button” but now these expectations will be brought into clearer focus. Will this be possible with Azure AI or Copilot? – time will tell.
In short, work volumes, profiles and expectations will alter. Those law firms that succeed will be those that move and adapt with these trends and launch new products and services with new approaches.
2. Law Firms – Generative AI will give rise to material changes in law firms. Goldman Sachs and others have quoted some fairly startling statistics about how law will be affected more than other sectors, albeit McKinsey are more cautious. For law firms this will have several dimensions:
a. There will be a need to look at the whole production lifecycle against a risk matrix. What areas of work can and can’t be replaced?; what areas of work are expensive admin?; where can documentation be produced more quickly?; what will the ideal staffing profile of a law firm need to look like? Theoretically junior resource won’t be needed as much but that isn’t really possible and appropriate for a firm that wants to be sustainable to develop up and coming talent. What new areas of work will emerge?
One law firm partner of a large firm said to me that there is likely to be an overhead challenge (i.e. they recognise that certain areas of work could be classed as very expensive admin and could be replaced but the challenge for the law firm was how quickly they could replace these areas of work given their overhead and cash in/cash out model).
There is also a huge repositioning piece. If the nature of incoming demand alters, how will a law firm adapt and reposition itself? There is a huge opportunity here but it requires thought and agility. What does the future law firm look like and how does a firm get there?
b. An area law firms need to think carefully about is insurance and risk. There are many players involved in Generative AI and the SaaS world but when looking at their terms and conditions it is clearly apparent how very few take on any material liability for their outputs. The expectation of law firms is hugely different (i.e. they are held to account and are sued if they get things wrong) and so how do they navigate to a position with Gen Ai where they are both leveraging the technology and adapting what they do but are being paid commensurately for taking the levels of risk that they do?
c. From an insurance perspective, what we are hearing from insurers is that they are happy with law firms using generative AI but they want all work to be supervised and see clear evidence of supervision on their file. Again, this shows a change in emphasis and law firms need to adapt (even in relation to what they keep on their files). Also, does proper supervision mean reading any document from scratch? If so, this will be time consuming (compared with say document automation). Or will this be where Generative AI verification or playbook tools come into their own?
d. There is obviously a huge opportunity for law firms to release products and some of these could potentially be very remunerative if delivered on an annual recurring revenue basis. This is not for the faint hearted though as the model and investment profile is vastly different. Also, this is an area where some law firms have less experience.
e. The hourly rate is brought into focus as well. Some say it will die in that jobs will become task based; others are talking about adopting different pricing for a human service. One thing we are hearing from clients is an increasing desire for law firms to be transparent about where they are and aren’t using generative AI and what is happening with their data. What they are already saying is they don’t want to be paying for a human service that is delivered in a fraction of the time with generative AI. The desire for transparency seems to be increasing.
3. People – It is really important to engage with your people in relation to generative AI. There will be some who will be scared whereas others who will see this as a career defining moment where they can reinvent themselves. Jobs will go and new jobs will be reinvented. Everybody needs to share their strategy and engage as people will want to know your strategy. As a minimum we will see new roles emerge such as prompt managers (albeit this is being increasingly built into the front end of Generative AI tools). In some areas, admin work will disappear.
It is also important to revisit how we train people. This has many dimensions including:
a. Training people on Generative AI (e.g. what to do and what not to do, the guardrails).
b. Educating them on the whole world of legal operations and looking at things differently (e.g. The Professional Alternative - The Professional Alternative is one way of doing this).
c. Training people on some of the areas of work where they will no longer learn from on the job experience.
d. Training and risk management. This is really an area where the rule book is being rewritten.. This is also an important area to monitor carefully. As a very good friend of mine says, “you need to treat Generative AI as a very junior enthusiastic assistant whose work you need to supervise carefully”. For senior people, Generative AI can be a real game changer. For more junior people however, they may find it hard to know what is correct and incorrect and may find it hard to know when a tool is hallucinating. Their work will in some ways be harder to supervise as it may be less precedent based (albeit there are ways of addressing this) with no/little data lineage as to how the content has been derived. It is also well documented how some people may treat this as an opportunity to represent that they are better or harder working than they actually are. All of these areas need to be addressed and supervision will be harder and will need to be done differently.
4. Culture – Generative AI will be a real test of the culture of an organisation. We are likely to go through unprecedented change. In my experience this will be where you see the true culture of an organisation and where both good and bad behaviours can emerge.
To state the obvious, law is going to become a much more tech and data driven industry . We are probably going to see a new breed of partners, lawyers and assistants emerge. The people who have been the stars and role models in the past (e.g. for being hard working or rainmakers) may not retain the status going forward and may be replaced by others as the job will be different. It is much more likely we are going to be in a much faster data driven world. I think many outside the profession do not always appreciate how hard a lawyer’s job already is but if we amplify the speed and increase the risk of the type of work that they are doing this may become much harder.
A number of firms are looking at potentially reorganising how they are set up, restructuring themselves better to manage this change.
5. Quality – We have had lots of discussions about quality. On the one hand quality may improve (i.e. if an inexperienced lawyer or client produces a piece of work the output may be better than it would have been). The challenge is that perhaps this may happen in an area where they would not have sufficient skills to recognise errors. From a more negative perspective we also have to recognise there is going to be a starburst effect where weaker content is produced on the back of a corpus of weak content. People may focus Generative AI on flawed content and negotiated drafts or drafts based on the specific circumstances of a deal. Looking to a positive though, Generative AI is a great risk management tool, e.g. a prompt on what are the top 25 points we should look at when reviewing a joint venture agreement, provides some pretty credible answers.
There are lots of issues which may distract from quality. We have already talked about hallucinations and the corpus of content perhaps not being as good as it ought to be, with no data lineage. We also have issues such as people not yet being used to drafting prompts thoroughly. Sometimes, when you ask for an agreement to be produced it won’t ask you if it is vendor or purchaser biased for example which makes a big difference. Many people think the answer to this is leveraging content on Azure or leveraging content on Co-Pilot. Again, we need to take great care. Any current, former or practising lawyer (like myself) will know only too well that on a deal there may be multiple drafts. The first draft may have been shaped to deal with particular circumstances, e.g. bargaining position or history. The last draft and everything in between will be negotiated. The reality is none of these drafts will be completely perfect to point Generative AI at and perhaps this brings you into a world where we have to continue to recognise the value of precedents.
Increasingly, our data quality and knowing how our data is structured will be key to the work we do. To state the obvious we are also going to see some challenges on confidentiality. These could be on a simple basis of people putting confidential information into say one of the public large language models. On other occasions we are going to see data leveraged that shouldn’t be. Many argue there are pros and cons of existing document and email management approaches but one thing that is apparent is that these systems have put a lot of effort into segregating confidential and price sensitive information and it is important that Generative AI respects these boundaries. Work is being done in this area but at the time of writing this article it is not complete. This is a material issue given that the bulk of law firms’ work content is stored in these systems.
6. Competition – I and others firmly believe that Generative AI is not just “another technology”, it goes to the heart of a lot of what lawyers do (i.e. they generate documents, analyse them and provide advice). We see some firms that will really succeed in this space and others who will perhaps fail. There will also be underlying issues behind this such as data quality, culture, speed of reaction. We already have a number of AI providers in the market and some of these may struggle. This is not because they are bad tools but because law firms can’t do everything and in fairness to Microsoft and Azure and the other large vendors in this space they are making things very easy to embrace at comparatively low cost. Many of the emerging tools are much more “oven ready”, meaning that they could be leveraged quickly as are the embedded AI in playbook tools. We are going to see a whole range of new products and also new services such as from Thomson Reuters who are already connecting their products with Co-Pilot. Our competitors of the future will alter and dare I say it, this could be a huge moment for in-house legal teams to really drive efficiency. That in itself could alter the balance and quantum of work. We are working on a number of projects in this space and the level of ambition we are seeing is staggering.
Conclusion
We and many of our contacts believe Generative AI is going to have a material effect on the legal market. To understand this we need to scenario plan but importantly when looking at this there are a wide range of dimensions to consider and so hopefully this article will help. What is important to remember though is that this area is evolving really quickly. The pace of change is huge and so doing nothing is absolutely not an option. Some ideas will work and others will fail but no one will come up with a strategy which is 100% correct. We all need to engage and adapt. We need to experiment and make sure we are not left behind.
One other key point worth bearing in mind is that Generative AI (and to be fair to Microsoft 365) is a great leveller - Microsoft 365 “The Great Leveller” - Managing Partners Take Note — Hyperscale Group Limited). By this I mean, in the past your technological capability was linked to the size of your wallet but this is changing. Every person, firm or in-house team with a good internet connection in many ways can have the same Gen AI capability as the biggest law firms or legal functions in the world. Never before have I seen this in my career and it will be interesting to watch it unfold.
Hyperscale Group are an Independent Digital, Innovation and Operational Advisory and Implementation Business with over 25 years plus deep market experience. We work for In-House Legal teams and Professional Services Firms all around the world and support them developing and implementing their strategies. We help our clients to make the right things happen.
Our Experience: Click here - Our Services: Click here - Client Testimonials: Click here
How We Work: Click here
For more information, please contact Dereksouthall@hyperscalegroup.com
