There are two areas that have become particularly problematic in the use of Generative AI (GenAI) and Large Language Models (LLMs) in business: it's use in legal cases by attorneys and the destruction of the talent pipeline across many industries.
First, the problematic use in the justice system. For some reason, this hasn't just been just a few outlier cases. There are numerous examples where a brief or submission to the court has been caught by a judge with AI hallucinations.
The core of the problem is attorneys using AI to do their jobs for them rather than assist them in their workload and potentially save a little time.
At worst, though, is the concept of generated evidence. Unscrupulous attorneys (or other parties) could use AI to generate deepfakes of images, videos, audio, and more to either frame someone or convince a jury or judge that a guilty person is innocent. Our legal system and our culture is just not ready to handle the level of fake evidence that AI can create. We prove this daily as people continue to click on phishing emails that pretend to be from trusted sources.
For now, though, we have seen numerous examples of attorneys creating briefs using AI and not bothering to go back and check their work. This is a problem in many industries but most noticeable here as the legal world forms the foundation of our society. Unfortunately, it has grown to rely on the bloviated and algorithmic submissions of parties in text format. This is partly due to the spider web of precedent that is our legal system.
When attorneys use AI to generate a "brief" and do no review or editing before submitting, they are abandoning the integrity of the system for laziness or money. They do us all a disservice as the justice system is at the core of how and why our society functions.
How should it be done? Using AI to generate your legal briefs isn't the problem. Just review the documents and meticulously fact check every word before submission. Why is this hard? Because it doesn't fulfill the massive "efficiency boost" promised by the AI hype bros. Why did you trust those snake oil salesmen in the first place?
The next area, and one that is closer to us here in the tech world is the gutting of junior and entry-level roles based on empty promises that AI will handle all of those mundane tasks. Try looking for an entry-level role in tech right now, you'll cry.
Sure, we should be looking to AI to take over some of our daily drudgery but not at the expense of allowing juniors to enter the career field and grow into mid-levels and seniors. There is nothing wrong with standing on the shoulders of giants so long as we do it wisely.
However, when we run out of qualified professionals because there was no path into the career field, then it becomes a problem. We don't instantly mint senior professionals without people going through junior roles and years of learning and doing first.
This is a very short-sighted action by (probably) someone who only manages from a spreadsheet. Our economy has that problem systemically, though. We rarely consider the long-term ramifications of our decisions.
What can be done? Well, two things really. First, let's not just instantly decimate all hiring pipelines. You may make a quick buck but you and your industry will suffer in the long-term. People make decisions about career fields based on opportunities available. Continue to hire those entry level folks out of college and other training programs so they see that jobs are available. And also, so you have those seniors in 5-10 years.
Next, take a hard look at the baselines for each skill level in all your job roles. Explore how you can elevate the expectations and output without raising the experience requirement to entry. This shouldn't be solely based on the hyped up promises of AI but also look at other ways to automate tasks. Look for ways to cut out the unnecessary tasks from those roles to allow the person to accomplish bigger and better things for you.
While AI may not seem like a cybersecurity thing, it is. There are numerous cyber risk implications from the use of AI in your business, from simple to complex. There are also, as stated above, several areas that just make good business sense not to dive in head first and consider more long-term implications. Need help working through when and where you'll use AI in your business? Call us.
References:
https://cyberscoop.com/ai-deepfakes-causing-big-problems-in-the-legal-sector-aba-report/
https://www.darkreading.com/cybersecurity-careers/with-ai-reshaping-entry-level-cyber-what-happens-to-the-security-talent-pipeline-
