.png)
The future is here, as more companies embrace artificial intelligence (AI) to help them sort and analyze data faster than ever before.
However, rather than witnessing a hostile robot takeover from Sci-Fi movies of the 20th century, organizations across industries recognize that AI can’t replace human thought or intelligence.
Instead of being used as a replacement for human judgment, AI should be used as a supplement to critical thought. This is often referred to as human-in-the-loop machine learning, which integrates human intelligence with artificial intelligence to create a feedback loop that improves AI by teaching it how to achieve the desired output.
In legal review, keeping humans in the loop is critical for ensuring that the integrity of the law is upheld. When it comes to redlines, using assisted review that combines human and artificial intelligence is more effective than taking a fully automated approach.
AI in Legal Review
In legal review, it's easy to be lured by automated redlining software, which is supposed to save editors time by scanning documents to find and redline errors. Powered by AI, these software programs claim to have the ability to read and edit documents more accurately and efficiently than humans can.
The concept of automated redlines is enticing for busy professionals who don't have any time to waste. Who wouldn't want to turn over a manual task like redlining contracts to a machine that alleges to be able to do the job better?
Unfortunately, as the saying goes, if it sounds too good to be true, it probably is.
AI poses many challenges in the nuanced and technical realm of contract negotiation. By turning your redlines over to a machine, you risk having more inaccuracies and less clarity in your contract. Additionally, the process can end up costing you valuable time instead of saving it.
Understanding AI's Limitations
AI systems replicate how the human brain works. They are fed data by humans and learn to recognize patterns. As the machines see more data, they become more intelligent and better able to adapt. This ability to learn and change behavior based on new knowledge is what makes AI-powered machines different from other types of computers.
Nobody's arguing that AI isn't powerful. It has the potential to change the way we work by taking over several job functions that humans previously performed. AI chatbots, for instance, can deliver frequently asked information to customers. Questions the AI cannot answer can be routed to a human, who now has more time to spend solving customers' complex needs.
The challenge is that AI's prevalence in our lives has caused many people to assume that it is a perfect type of machine. In reality, though, AI is not as intelligent as humans. It mimics human behavior, but it can never fully replicate it. While AI can gain knowledge, it can never gain specific cognitive abilities that will make it more valuable than a human.
Machine learning (the engineering field underlying AI) was never intended to replace human thoughts and actions. Instead, ML was designed to augment human behavior by working side-by-side with humans to analyze data. Some areas require more human intelligence than others, including contract negotiation.
The Power of Human-Assisted AI
The term human-assisted AI describes how humans interact with AI technology to make informed decisions. Most managers (87%) believe the future of human-machine collaboration will rely on a hybrid model that combines human judgment with AI-informed data.
A recent study from MIT Sloan found that the human filter that acts as a liaison between AI and an organization is what makes the difference when it comes to making organizational decisions based on AI. Their data analysis found that individuals don’t make the same decisions when presented with the same AI-informed data. Instead, individuals make very different choices based on their personal preferences, biases, and experiences.
Critically, MIT Sloan's study found that executives' AI-based decisions directly affect the organization’s financial performance. So, organizations need to recognize that there will always be people making decisions, even when they use data from AI. Because of this, organizations need to ensure their executive and leadership teams can make the right decisions based on AI inputs.
Challenges of Using AI to Redline Contracts
While AI offers many practical applications that benefit human laborers, we need to recognize that it cannot replace humans when critically thinking through essential tasks. Adopting an automated process for contract negotiations can save time and reduce human error for legal professionals. However, it is also necessary to keep humans in the loop to ensure that the law is correctly applied.
Legal professionals face several challenges when using automated redlining software for contract review.
First, contract negotiations are constantly evolving, making it difficult for markup to be automated based on new information.
Next, contracts are too semantic to get them right reliably. AI software often operates based on a strict set of rules that may not result in the correct rhetoric in a contract.
Finally, it takes a significant amount more work to correct AI's output than just to do the redlines yourself. Most of the time, it's not worth the additional effort to use AI software than it would be to review and edit contracts manually.
AI cannot keep up with always-evolving negotiations.
The world of contracts is constantly changing in unexpected ways that AI cannot anticipate. For example, a pandemic has now been added to force majeure following COVID-19. AI software didn't know to correct for this when the change happened. Depending on the software, it might still not know.
AI does not get information about industry changes as quickly as the humans responsible for the language in the contracts. It also does not understand the nuances between industries enough to make reliable recommendations.
Contracts are too semantic to get it right reliably.
Unlike humans who are heavily involved in contract negotiations, AI software is not privy to the full context of a contract that can influence its semantics. It can obscure meaning by being more concerned with specific "rules" than with presenting information with clarity and precision. Therefore, it is unfair for legal professionals who use redlining software to expect it to be right 100 percent of the time.
Think about the last high-risk contract you redlined. Would you have trusted a machine to identify and understand the context behind every word in the contract n a way that gave you accurate redlines?
It takes more work to correct AI's output than to do it yourself.
If you've ever attempted to redline with AI software, you know how time-consuming it can be to go through the AI's edits on top of your own. Track changes can quickly turn into a nightmare as you sort through all the redlines to determine which ones are valid and which ones need to go.
No matter how well your AI software understands your algorithms, you will still need to employ a human reader to review its redlines. In some cases, this could save you time—if the redlines are accurate and correct. If they are anything other than perfect, though, whichever unlucky human ends up reviewing the AI's redlines will likely waste a significant portion of their day.
Assisted Review is A Better Way to Redline Contracts
The challenges listed above don’t mean that legal teams should give up on incorporating AI into their contract negotiations altogether. Instead, they indicate that lawyers should exercise caution when redlining with AI to ensure that the outcome is accurate and fair for all parties involved.
In contracts, using precise language is critical. You can't afford to waste time fixing an AI's redlines or risk signing a contract that is full of inaccuracies. AI cannot replace the need for negotiation and advocacy by humans. It might be able to check for grammatical errors, but it cannot understand nuances in language that can completely change the meaning of a sentence.
Another way to think about automated review is in terms of baseball. Fully automated review gets you to first or second base (albeit blindly). You can cruise to first or second base without even breaking a sweat when you rely solely on fully automated review.
However, if you want to get past second base, you still have to review it manually. Often, you need to review it multiple times before you can get to third base.
Assisted review is what allows you to round third base and make it home.
All of the challenges legal professionals face when using AI to redline contracts are frustrating. Fortunately, there's an alternative that doesn't require you to return to old-school methods of using track changes in Word documents.
See for yourself why assisted review is 83 percent more effective than fully automated redlines. Book a demo with DocJuris today.
