Someone who has had a tough day at the office might be inclined to (on occasion) envy the life of the family dog. All that ’ol Rexi-boy really has to worry about is: 1) dinner time, 2) playtime, and 3) sleep time. Not much else gets in the way.
Not a bad alternative to stress.
However, you’re not alone, dogs are now seeing their day in court, too. Namely, dogs that are created based on AI technology. They are being saddled with responsibility and performance metrics.
And legal considerations.
As reported in a special CNBC report1, MIT has developed an incredibly agile robotic dog that uses AI technology. It is currently being fine-tuned to run an obstacle course. And private firms joining the dog-dev endeavor as well, including Boston Dynamics. Their AI-K9 is used in the energy sector to rover through an electrical-grid facility and monitor the equipment in energized areas where humans cannot go.
Dogs have become autonomous machines.
But what happens when one of those Rovers does something bad… as in, “The dog ate my homework?”
Say the Boston Dynamics dog removes some debris it finds during its route. But, that debris just happens to be an important report the foreman set on the ground for a moment.
Can we yell at the AI for that? “Bad dog!”
Who can we blame? When machines are trained to do reasonable things but some unexpected results may occur, who is liable?
In a classic Gary Larson “Far Side” cartoon that personifies animals as humans,
a bespectacled “schoolmarm” (adult dog) chastises her entire classful of young student “pupils” (puppy dogs)…
demanding to know which ones “did not” eat their homework on the way to class that morning.*
How AI Liability Becomes a Challenge
Chief Executive magazine2 reports that, in general, in tort law, humans are legally liable if there is something they could “reasonably” anticipate being a negligent issue… but fail to do something about it. For example, an office-building owner should clear the ice or debris from a loading dock where delivery people regularly walk back and forth. Otherwise, they might slip and get hurt. Pretty obvious. Pretty reasonable. Keep it ship-shape or you’re liable.
But in the energy-sector example above, the dog was reasonably trained to remove debris, and it did just that when it “ate” the report. So, it is not to blame. However, in tort law, a plaintiff attorney may argue that it should have been “reasonably” trained not to include sensitive documents in its clean-up duties. Meaning, that would fall back onto the AI developer (and maybe others who approved the programming, such as the AI customer).
AI changes the way liability is viewed and AI creators may be the ones who liability comes chasing after. But AI users might become liable, too. It’s a new legal frontier.
How AI Liability Remains Manageable
As companies, including marketing agencies, proceed along the path of developing AI and SaaS, and digital tools, they need to be cognizant of unexpected outcomes from the intellectual property being created. For example, maybe an app the agency creates results in a loss for another party.
“As risk ratchets up, more (caution) is better than less,” according to Sharon Toerek. She is an intellectual property and marketing law attorney, with a national firm based in Cleveland: Legal + Creative. She helps marketing agencies protect their intelligence. And themselves.
“Keep business foundations intact. It’s intellectual property, so add appropriate new IP-related language to Master Service Agreements. And consult with your attorney, tax advisor, and insurance broker (errors and omissions, liability, etc.),” she says. “The amount of premium increase to cover yourself is very manageable in comparison.”
Overall, AI is creating new issues and possible outcomes that are not totally clear. And it may always be an evolving area of development. So as things become less clear, make the agency’s position on liability more clear.
LEARNING ABOUT AI?
BONUS MATERIAL: AI RESOURCES… Quick Reads, Quick Links…
From Analytics Insights magazine online, check out basic definitions and terminology related to AI and machine learning. Many helpful links within this article and within this site… click around for a wealth of objective info.
Find out more about generative apps for creative text and artwork, such as ChatGPT, LensaAI, Midjourney, and DALL-E 2. Other related articles link within this one.
At online educator Emeritus.org read about AI business applications by industry, such as how an airline can predict exactly how many flights it needs to get its passengers where they want to go… while keeping overbooking or underbooking in check (and many more examples).
*For more dog humor, check out Wiener Dog Art: A Far Side Collection, by Gary Larson, available in stores and online along with other classic collections such as The Far Side and The Far Side Gallery.
- M. Petrova, “The rise of the robotic working dog,” CNBC Work, https://www.cnbc.com/video/2021/12/26/robot-dogs-from-boston-dynamics-anybotics-ghost-robotics-start-work.html, Dec 26, 2021.
- D. Fisher, “AI Tests Tort Law,” Chief Executive, https://chiefexecutive.net/ai-tests-tort-law/, data pulled Jan 25, 2023.