Share It
There are many stories out there about the amazing things that organizations can already do with generative AI technology. But for every success story, such as the one about GPT-4 passing the bar exam in the 90th percentile, there is a hallucination story such as the Mata v. Avianca case, where the lawyers issued a filing with fake ChatGPT-generated case citations.
We have so much to learn about generative AI technology, yet, while we’re trying to learn about it, the technology is continually changing and evolving. Things are moving so fast that we don’t know what new features to expect and whether they could lead to other potential challenges. As the stakes are higher than ever, corporations must be cautious when it comes to building workflows and processes around current iterations of generative AI, when the risks of those current iterations aren’t fully understood and when newer iterations and solutions may come along at any time that could change the game dramatically.
Are you prepared for the wild ride associated with generative AI’s rapid evolution? In this post, we’ll look at why it pays to be careful when it comes to innovative solutions such as generative AI, and how your organization can implement a cautious approach to generative AI that balances the potential for innovation and growth with the need to remain flexible and effectively manage risks.
Innovation Often Doesn’t Mean Instant Success
History is replete with examples of innovations that ultimately became highly successful, but only after they matured significantly. Here are three examples:
The Internet
The internet is considered to have originated as far back as 1989, when a new protocol — HTTP (HyperText Transfer Protocol) — was built over the existing TCP and IP protocols. However, the internet was initially a tool used by scientists and academics to share information. Its potential for commercial and personal use wasn’t immediately recognized and the technology took years to evolve from communication via dial-up modems to the high-speed connections we all take for granted today. Now, the internet is foundational to modern society, enabling global communication, commerce, and access to information.
Mobile Phones
Mobile phones go at least as far back as 1984, when the Motorola DynaTAC 8000X (known by many as the “brick phone”) came onto the market with a hefty price tag of $3,995! Mobile phones evolved considerably for more than 20 years until the first iPhone was released in 2007 (followed by the Android operating system in 2008), which reinvented the mobile phone into the current generation of smartphones popular today that are indispensable to our lives.
Electric Cars
You might think electric cars are a more recent invention, but the first electric vehicles (EVs) were invented all the way back in the 1800’s! Over the years of the 1900s, the popularity of EVs was overrun by gas powered vehicles which could travel much further and the use of EVs became limited to isolated uses, such as golf carts. However, that began to change in the 2000s, as car makers such as Tesla began to offer all-electric cars that could travel longer distances. Today, with advancements in battery technology and a growing focus on sustainability, EVs are becoming increasingly popular and are seen as the future of transportation.
These are just a few examples of innovations that have become integral to our way of life today, but which took a long time to evolve and mature to do so. Like these innovations once were, the application of generative AI models to business use cases, especially eDiscovery, is in its infancy. While the potential of generative AI technology is tremendous, the AI models — and how we use them — will change dramatically over the next several years and beyond.
A Cautious Approach to Embracing Generative AI
That doesn’t mean that you should wait to embrace an emerging technology such as generative AI. Corporations can do so in a manner that involves balancing the potential for innovation and growth with the need to maintain flexibility, while managing risks and ethical considerations. Here are some considerations for a cautious approach to embracing generative AI:
1. Assess Ethical Implications
Consider the ethical implications of deploying generative AI, including privacy concerns and the potential for bias in AI-generated content. If your organization doesn’t already have a framework or set of guidelines for ethical AI use, now’s the time to develop one.
2. Learn From Previous Successes with AI
As we discussed in this blog post, there are several examples of proven disruptive AI technology for eDiscovery around which your organization has likely already built productive and efficient processes and workflows (which probably evolved as those technical capabilities matured). Keep those in mind and learn from those successes as you develop workflows to support generative AI capabilities.
3. Promote Transparency and Accountability
Be transparent about the use of generative AI in your workflows and implement mechanisms for accountability, ensuring that there are processes in place to review and address any issues that arise from AI-generated outcomes. “Trust but verify” should be a standard mantra for the use of generative AI, until the technology (and best practices for use of it) matures considerably.
4. Start Small and Scale Gradually
Begin with pilot projects or small-scale implementations to assess generative AI’s impact and effectiveness, starting with a focus on use cases that minimize risk (such as document summaries and first pass review, as opposed to responsiveness review, where the defensibility standard will be considerably higher) and test runs on data sets with known outcomes to verify results.
5. Stay Informed
Most importantly, continuously educate yourself and your team on the latest developments, capabilities, and limitations of generative AI. Understanding the technology’s current state and potential future directions can help in making informed decisions regarding how you should leverage new capabilities and how doing so will impact current processes and workflows.
“Trust but Verify” With Gen AI
Generative AI technology has great potential to revolutionize how discovery and investigations are conducted; however, the technology is still in its early stages and is just beginning to be applied to eDiscovery use cases. That’s why it’s great to be proactive, but you should do so with caution.
It’s important to have a set of ethical guidelines for the use of generative AI, “trust but verify” in your organization’s use of the technology, remain flexible regarding likely frequent adjustments to your processes and workflows and stay informed as the technology rapidly evolves. Doing so will enable your corporate team to maximize the benefits of this amazing technology (now, and in the future) while minimizing the risks associated with it.
For more on Casepoint’s GenAI principles, click here.
