The Legal Hazards in Using GPT Text Generated Artificial Intelligence to Write Contracts

One of the key selling points for using the emerging artificial intelligence platforms such as GPT3 in the commercial context is that it can write text generated business contracts for its users. While businesses have always had access to legal forms to use as a template in such situations, this is different. Artificial intelligence will provide a user with the ability to describe the nature and parameters of the transaction (“I need a contract for a commercial lease of warehouse space in the building next to my current address in which I will warehouse canned food with 24/7 access using a keypad code”) with the result that the final product will immediately arrive at the user’s digital doorstep, ready to be signed.

Sounds good, but there is more to the analysis.

Can Artificial Intelligence Platforms Actually Write Commercial Contracts Customized for the User’s Business Needs?

The answer to the above question is a resounding no. Let’s say you ask GPT3 to provide a lease for the commercial warehouse referenced in the above query. Admittedly, it will be able to perform such functions as filling in the blank fields with the parties’ identifying information and including the term and price for the rental. Likewise, it may even be able to take into account the fact that you are the tenant, and therefore draft the contract in a manner that is tenant-friendly.

For example, this might include such items as:

• how close to the end of the term a notice of termination must be given by the tenant to avoid an automatic renewal of the lease;
• whether the requirement that the tenant indemnify the landlord by compensating the landlord for the damages caused by the tenant’s acts will include reimbursement for the landlord’s attorney’s fees; and
• the extent of the tenant’s joint and several responsibility for environmental contamination at the warehouse that is discovered after they vacate.

The Main Problem is That Artificial Intelligence Applications Cannot Readily Discern Between Different Versions of Text Generated, User-Friendly Contract Provisions.

In each of the above examples, the artificial intelligence application can provide a tenant-friendly set of lease provisions, but it cannot discern between the dozens of variations of those sorts of provisions. In other words, artificial intelligence enhanced platforms, such as adjuncts to, or variations of, the nascent Google platform LaMDA or Microsoft’s ChatGPT powered Bing, may be able to create a provision that is market standard in that part of the country for that sort of warehouse storing those sorts of products. But what it cannot do is have the judgment to pick a provision that would differentiate among a range of negotiating positions to ask for something unusual to which the landlord might agree. That would depend upon the underlying circumstances, something that ChatGPT would not necessarily know. Examples of this include if the landlord were in the unfortunate situation of having three tenants terminate their leases at once or if the landlord was trying to sell the warehouse that was being rented. In other words, the application can come up with an admittedly correct tenant-friendly provision; but it cannot employ the judgment necessary to assess which of a range of such provisions will have a likelihood of working under the circumstances.

Some examples of the Limits of Using Artificial Intelligence to Draft Contract Provisions

Here are some examples of a wide range of tenant-friendly lease provisions, each of which would be correct, but all of which would provide different rights and responsibilities among the parties.

• LEGAL ISSUE: How close to the end of the term must the tenant give notice of termination to avoid the lease automatically renewing?
  • Variation #1 of the Lease Provision: The Tenant must give 90 days notice of non-renewal or the lease will automatically renew.
  • Variation #2 of the Lease Provision: The Tenant must give 90 days notice of non-renewal or the lease will automatically renew. However, if the Tenant misses that deadline, they may still give such notice of non-renewal up until the end of the term. But if they do so, they will need to pay the Landlord an additional three month’s rent as a penalty.

• LEGAL ISSUE: Will the Tenant be required to indemnify the Landlord for the Landlord’s attorney’s fees that were incurred by the Landlord because of the Tenant’s acts or omissions?
  • Variation #1 of the Lease Provision: The Tenant shall hold harmless, indemnify and defend Landlord for any claims by third parties based upon the acts or omissions caused by the Tenant. However, the obligation to defend shall be limited to reimbursing the reasonable attorney’s fees and costs of the Landlord in that regard.
  • Variation #2 of the Lease Provision: The Tenant shall hold harmless, indemnify and defend Landlord for any claims by third parties based upon the acts or omissions caused by the Tenant. However, the obligation to defend shall be limited to reimbursing the reasonable attorney’s fees and costs of the Landlord in that regard, and the Tenant’s responsibility for the Landlord’s attorney’s fees and costs shall be capped at $10,000.00.

• LEGAL ISSUE: What will be the extent of the tenant’s joint and several responsibility for environmental contamination at the warehouse that is discovered after they vacate the premises?
  • Variation #1 of the Lease Provision: The Tenant shall be jointly liable solely for its proportionate share of the damages caused by contamination at the warehouse that can be reasonably proven to have been caused by the Tenant during its tenancy.
  • Variation #2 of the Lease Provision: The Tenant shall be jointly liable solely for its proportionate share of the damages caused by contamination at the warehouse that can be reasonably proven to have been caused by the Tenant during its tenancy. However, Tenant’s liability shall be limited to one and one-half the total amount of Tenant’s annual rent remitted during the last twelve months of its tenancy.

GPT and Other Artificial Intelligence Applications Cannot be Relied Upon to Negotiate Contractual Terms with Sufficient Judgment and Reasoning.

Essentially, GPT and other text generating artificial intelligence powered applications can correctly draft a contract, but not choose wisely among multiple ways of writing the contract correctly. The point of having experienced counsel negotiate a business transaction is to utilize nuance and a sense of the other side to maximize their client’s advantage. That takes perception, emotional intelligence and gut-instinct: qualities that are lacking (so far) in artificial intelligence applications.

The Nissenbaum Law Group welcomes potential clients who are seeking legal advice in utilizing artificial intelligence generated content for their businesses.

The Nissenbaum Law Group has a focus on digital law that includes representing businesses that seek to take advantage of cutting-edge technology, such as artificial intelligence enhanced platforms, including adjuncts to, or variations of, the nascent Google platform LaMDA or Microsoft’s ChatGPT powered Bing. The firm encourages any businesses with questions or concerns about the legal issues engendered by using artificial intelligence platforms—including the copyright and other intellectual property protections that may be necessary— to contact Mr. Nissenbaum at gdn@gdnlaw.com.