This week’s blog post centres on a question in BJL’s FAQ section:
Q: Why would a client pay me to use this system when they could pay for access themselves and get their own answers?
The answer provided on the website focuses on how professionals bring to bear their own experience and judgment to the predictions made by Tax Foresight and Employment Foresight:
A: Using Tax Foresight and Employment Foresight appropriately requires exercising professional judgment in presenting information to the system. Particular Tax Foresight and Employment Foresight results are based on the information provided by the user. Moreover, professionals bring to bear their own experience and judgment to the predictions made by Tax Foresight and Employment Foresight, which would be missing if a client relied on Tax Foresight and Employment Foresight directly. For these reasons, Tax Foresight and Employment Foresight are currently being marketed directly to tax and employment professionals.
This week we’re explaining exactly why Blue J Legal’s software is a technology specifically aimed at helping, not replacing, professionals.
The conversation around artificial intelligence often focuses on how AI can replace people, not how it can help them. This can create a negative perception of AI and AI-powered technologies. I spoke with Avi Brudner, Blue J Legal’s COO, about Tax Foresight and Employment Foresight, and the difference between legal advice and legal information, which is crucial to understanding what role BJL’s products play in practice.
“Our system isn’t meant to provide legal advice, but rather it’s a legal research tool,” says Brudner. “We’re essentially providing the best legal information to professionals so that they can provide the best advice to their clients.”
BJL’s products benefit professionals by doing heavy lifting, research-wise, that gives professionals data-backed predictions that can be used as jumping off points for further research, helping users of the software get to the next step faster, and more efficiently.
WHAT DO YOU GET?
But what does this heavy lifting look like? First, users are asked to fill in a short questionnaire. From there, the ‘answer’ comes in the form of a succinct answer, percentage confidence, a one-page explanation, similar cases and references.
The percentage confidence measures how confident the system is in the answer, expressing this confidence as a percentage based on the data from the questionnaire and relevant past case outcomes.
The explanation is a one-pager that explains “how certain factors may have interacted in a user’s scenario,” says Brudner. “It is generated automatically by combining a user’s responses to the questions with the machine learning algorithms driving the analysis. The explanation helps to interpret the directionality and impact of certain factors.”
The similar cases are matched according to the factors that were inputted in the questionnaire, serving as jumping off points for further research. These cases, says Brudner, “allow professionals to draw parallels and use those cases as appropriate precedents when advocating for clients.”
“We specify the result of each case very visibly in the list of similar cases,” he says. “So it’s really easy to see cases that are similar that went in favour of the client, or cases that are similar that are not in favour of your client.”
The references differ from the ‘similar cases’ in that they are cases that are frequently cited by courts in judgements related to the legal issue at hand.
PROFESSIONAL EXPERIENCE AND JUDGMENT NEEDED
The predictions that BJL’s software gives require the necessary context and background that a professional has to interpret properly and use to take the next steps. A certain level of understanding of the legal principles behind the legal issues is needed, says Brudner. Background knowledge is helpful before and after using the tool.
Even if someone who’s not a lawyer, like a potential client, uses this software to get feedback on their case, they’ll still need a professional to interpret the data and then help them by settling or taking the case further.
For example, an employment lawyer will know what to do when the AI says its 95 per cent confident in favour of a case, says Brudner. “So now what do we do? Do we write a demand letter to your employer? [Lawyers] will know the appropriate next steps.”
Brudner says that even given the results, professionals will likely want to do their own reading and confirmation that the given results are correct.
“We’re confident in the system, but professionals tend to want to do their own test to verify the predictions,” he says, following that this includes reviewing the explanation in detail and then reviewing the similar cases the system provides.
“A senior lawyer on their own has limited time and capacity to remember or research cases that are relevant to a client’s situation. The power of the system is that it allows you to consider all the cases at once. You benefit from the machine’s interpretation of hidden connections between the variables in all of the case law: you get a quantitative, data-driven prediction.”