Here Are the Practical Risks of Generative AI, Experts Say

Here are the Practical Risks of Generative AI Experts Say by Sarah Martinson

[Editor’s Note: Sarah Martinson, Senior Reporter, Legal Tech for Law360 covered the EDRM Flash Ripped from the Headlines Webinar on April 6. 2023. The webinar hosted AI experts and EDRM Global Advisory Council leaders Dr. Maura R. Grossman, attorney, special master and research professor at the University of Waterloo’s School of Computer Science, Dr. Johannes Scholtes, chief data scientist for EDRM Trusted Partner ,IPRO and John Tredennick, founder and CEO of Merlin Search Technologies, also an EDRM Trusted Partner. An excerpt from the article follows:]

When technologies like the printing press, steam engine or internet were introduced, people were also scared of them. I’m reminded of the old saying that I have to tell myself sometimes, ‘It’s never as bad as you think it is, or as good.’

John Tredennick

Prominent tech leaders recently signed a letter calling for a pause on the training of large language models that power generative artificial intelligence, warning that AI could lead to the “loss of control of our civilization,” but e-discovery experts said in a webinar Thursday that there are more practical risks.

Some of the practical risks of generative AI include lack of transparency; biased training data; outputs of false information, known as “hallucinations”; cybersecurity breaches and unintended consequences, panelists said during a virtual event hosted by e-discovery organization EDRM.

There are many more real-world problems that we need to be discussing.

Dr. Maura R. Grossman

Maura Grossman, a research professor at the University of Waterloo’s School of Computer Science, said she is not worried about OpenAI’s large language model, called Generative Pre-trained Transformer, or GPT, controlling the planet.

“There are many more real-world problems that we need to be discussing,” Grossman said.

We should put a framework around AI which follows our values and norms, and that’s exactly what OpenAI did with the reinforcement training” of GPT.

Dr. Johannes “Jan” Scholtes

Read the entire article here (registration required).

Listen to the webinar at your convenience here.



Author

  • Mary Mack is the CEO and Chief Legal Technologist for EDRM. Mary was the co-editor of the Thomson Reuters West Treatise, eDiscovery for Corporate Counsel for 10 years and the co-author of A Process of Illumination: the Practical Guide to Electronic Discovery. She holds the CISSP among her certifications.

    View all posts