Can you trust Artificial Intelligence for contract review?

In recent years there have been some excellent applications of AI technology, medicine is one that springs to mind. Machine learning algorithms have been hugely successful in the diagnosis of diseases. In some cases, these algorithms can be more successful than humans in these diagnoses. Admittedly the application so far is limited and requires large sample sets with accurate diagnoses in order to train the machine but once the accuracy approaches (or exceeds) that of what a human can do, then you begin to trust the output and the technology.

Reaching that level of trust is critical as without it you still need people to verify the output generated by the machine. Unless you can completely remove the people element with the application of AI technology can you really call it AI? This is the challenge facing many AI solutions today!

With the rapidly growing popularity of AI there is an increasing expectation upon software vendors to implement AI in their products. As soon as one vendor claims to have implemented others feel compelled to follow suit to stay competitive, but ask yourself this; has this application of AI helped solve the problem for my business? Am I still reviewing computer generated output from my AI solution?

When people do a better job than your current AI solution, then you are likely going to have to use people to verify ALL the computer-generated output. This can be very dull work, especially when it comes to applying AI to legal contracts as humans have to review the AI one document and data point at a time. Un-surprisingly the quality of the final output can be lower as when you had people involved in creating the output. AI is not ready to solve all problems and there are many domains where the technology is not sufficiently mature to benefit from its application.

At LIKEZERO we have observed these challenges in the contract digitisation space. We have found that to solve our client’s problems we have had to look beyond AI algorithms. If you want to learn more on how we have achieved this, please watch the following introduction to our approach.