ChatGPT and other forms of generative AI have been called a "game changer" for both the real estate industry and other industries more broadly. Certainly, since its inception in November 2022, ChatGPT has sparked worldwide attention and has emerged as a source of both excitement and trepidation for businesses.
The capabilities of generative AI like ChatGPT may be harnessed by the real estate industry to automate traditionally time-consuming tasks and to streamline processes to enhance the ways in which the real estate industry operates and responds to client needs. For public and private sector organisations in the real estate industry, generative AI holds the potential to (among other things):
- conduct property-based forecasts and calculations, such as rent calculations;
- assist with identifying the zoning and permitted use of certain sites;
- provide predictive analysis on property sales and trends, including to allow businesses to market their products and services to certain clients or in targeted areas;
- assist with due diligence activities undertaken prior to a sale or purchase of land through the acceleration of data collection and sorting;
- produce emails and other communications for clients, including to help personalise communications and improve the overall client experience; and
- reduce the time spent on property related enquiries by automating responses (for example, responses to tenant requests and enquiries).
While generative AI technologies hold great potential for experimentation and advancement, their use is not without risk. We have outlined below some considerations that businesses in the real estate sector may wish to consider.
1. Generative AI is not a replacement for human insight, review and oversight
While generative AI technologies, including ChatGPT, have immense capability, the information or responses produced are not always current and accurate. There are cautionary tales on the internet of where AI has gone awry or relied on "hallucinations" (in short, invented or fabricated data).
The use of generative AI should be viewed as a tool and a starting point. It is not the end solution. The real estate industry should ensure that generative AI inputs and outputs are reviewed and validated by an appropriately trained and experienced human prior to being inputted or relied upon.
There are several property-related tasks that require human input, review and oversight, including in respect of the customisation of contractual documentation and the informed analysis of the risks relating to a specific property transaction. Generative AI cannot replace informed (human) legal knowledge, expertise and experience.
2. Be wary of the legal implications of using generative AI
There are a range of legal considerations associated with the use of generative AI which need to be considered by entities prior to use, including as follows:
- Intellectual property (IP): Generative AI raises challenges from an Australian IP perspective, including how to grabble with concepts such as authorship. Traditionally, a work produced by technology is not capable of copyright protection under Australian law. However, this is an area of the law that is likely to develop and evolve in response to technologies such as ChatGPT. It is also possible that the responses and other outputs generated by ChatGPT and other generative AI systems might infringe the copyright and other intellectual property rights of third parties.
- Privacy and data risks: As noted in an earlier article in our series, businesses integrating proptech in their buildings or spaces need to adequately understand and manage the risks associated with the collection, use, disclosure and storage of personal information. This is especially important in the context of the use of generative AI which presents heightened privacy and data security risks. Once data is inputted into a generative AI system, there may be limited controls over how that data will be used and protected. In the absence of adequate controls, it will be difficult for entities to carry out their obligations under relevant privacy laws, including the Privacy Act 1998 (Cth).
In order to best protect data and to prevent a data or security breach, we recommend that information entered into systems like ChatGPT does not contain any personal information or confidential or sensitive client information. Businesses also need to be conscious of the "mosaic effect" when using systems such as ChatGPT. While one bit of data may not comprise any personal information of itself, when combined with other information that has been collected by the generative AI solution or which might be later inputted into the generative AI solution (including by other individuals or organisations), it might.
- Liability: Businesses should be alert to the liability risks associated with the use of generative AI. For most types of generative AI, there is limited to no scope for persons to individually negotiate specific legal terms and conditions, including to cover matters such as liability, indemnities, insurance, the protection of personal information and confidential information etc. This generally leaves entities relying on generative AI at their own risk and with limited contractual protection.
- Governance and policy compliance: Government agencies adopting AI need to ensure that they comply with any relevant government policies and procedures associated with its design, application and use. For example, subject to a limited category of exceptions, NSW government agencies are required to comply with the NSW AI Assurance Framework. The AI Assurance Framework assists NSW Government agencies to design, build and use AI-enabled solutions in a legally compliant and ethically conscious manner.
The legal environment and considerations are likely to evolve as governments around the world continue to grabble with, and seek to regulate, the use of generative AI. As at the date of publication of this Insight, in Australia, the Commonwealth Government is in the process of consulting with the public about the safe and responsible use of AI, including to help inform the regulation and policy response to the use of AI. The Department of Industry, Science and Resources has recently extended the closing date for submissions in response to this consultation to 4 August 2023.
As this is an evolving area of law, we recommend that legal advice is sought on a case-by-case basis.
3. Ethical considerations
The use of generative AI (and AI more broadly) raises ethical challenges which need to be navigated, including issues associated with data accuracy, integrity, reliability and potential bias and discrimination. Such ethical considerations are important to prevent the potential abuse or misuse of generative AI and to ensure that technology is not applied in a manner that may result in direct or indirect discrimination.