Well over half of corporate legal departments will have artificial intelligence systems for analysing and reviewing contracts in a year's time, a survey of global businesses has predicted. However it warns GCs to beware of claims that AI will work straight out of the box - or that it will end the need for review by humans.

Understanding Legal AI from the Inside Out, conducted by US analyst Ari Kaplan Advisors for Seal Software, a supplier of AI-based contract review systems, found that 37% of respondents already review contracts and agreements with the assistance of AI. Of those who do not, 47% say it is 'likely or highly likely' that their organisation will implement software in the coming year. This finding is likely to encourage a new wave of interest in an industry which is capitalising on new developments in machine learning and natural language processing to scan contracts and other documents in a fraction of the time taken by human experts.

According to the research, the main 'use case' for AI is the need to comply with data privacy legislation including the EU General Data Protection Regulation: 55% of respondents said they had implemented the system for this purpose. This was followed by the need to keep track of NDAs and procurement/vendor contracts. Brexit was mentioned by 27%. 

Respondents said that AI helped identify inconsistencies in their agreements, automates some elements of their review and helps manage mergers and acquisitions more effectively. 'We apply the technology during the negotiation stage to influence the discussion based on prior agreements and post-negotiations to ensure compliance with updated regulations, such as anti-money laundering laws and Brexit,' one lawyer is quoted as saying. 

The report also identifies an 'array of misconceptions' about contract AI. The most common is that it will work 'straight out of the box' - rather, organisations need to train both their team and the AI itself. Another misconception is that it will replace a human lawyer. 'There will still need to be human input and human review in the system,' one respondent said, 'probably forever.'