[00:00:02] Speaker A: Public Facing the podcast.
[00:00:10] Speaker B: Would you believe that AI can now generate deep fake videos right convincing scams and produce misinformation faster than humans can? Fact check it. And one major company just upgraded its biological risk assessment from medium to high.
And here's why this matters. Because the risks of AI, they aren't science fiction anymore. They're showing up in your inbox, your social feed, and at your workplace right now. Cue the intro.
[00:00:46] Speaker A: From the Natural State. This is Public Facing the podcast with Joe Holman.
[00:00:57] Speaker B: Hello and welcome. I'm so glad that you're here today. This is Public Facing and I'm Joe Holman. Now, thanks for joining us today for part four of our AI and Innovation series where we're tackling the topic that nobody wants to think about. But everybody needs to understand AI safety.
Be an active part of this conversation. Email us at helloblicfacingpod Stream Now. I'm not here to scare you. I'm here to help you navigate this landscape with your eyes wide open. Here's an uncomfortable truth, my friend. The same technology that can tutor your kids, help doctors diagnose diseases and write birthday cards, well, it can also be weaponized in ways we're just beginning to understand.
In today's roadmap, we're going to explore harm to individuals through fake content, examine broader societal risks from misinformation, and discuss what you can actually do to protect yourself and your loved ones.
Our main takeaway today is going to be to use informed engagement because that's the best defense. And fear without knowledge, well, that helps no one.
Now, AI enabled scams and manipulations, they're targeting individuals right now.
Look at deep fake harm. This is where non consensual intimate images and voice cloning for financial fraud and reputation sabotage. These aren't theoretical, they're documented and they're happening daily.
Now the voice clone scam. This is where a three second audio clip is now enough to convincingly replicate someone's voice. Imagine a family emergency call that sounds exactly like your child and or other family member.
And then there's AI powered phishing. That's where the generative AI is used to make personalized convincing scam emails that are scalable like never before.
The days of the obvious Nigerian prince emails, well those are fading fast. But the ones acting as say Publishers Clearinghouse Sweepstakes, well those are still circulating.
Now child safety concerns come into play. This is where the AI generated child sexual abuse material has become a growing problem that existing safeguards struggle to address.
Safety measures that were once effective Say a decade ago, are now being overwhelmed by the cunning, baffling and powerful swatches and schemes to reach little eyes and ears. The risk is high because the prey is precious. Our children.
You remember when we could spot a scam because of bad grammar? Yeah. Well, those days are over.
See, AI writes better English than most of us, which is both impressive and terrifying at the same time.
Now, personal vigilance matters.
Verify those unexpected requests through known channels, especially anything that's involving money or sensitive information.
So now we've covered the individual risks. Now I'd like us to take a look at the bigger picture.
Truth under siege. See, AI is supercharging the spread of misinformation and threatening our shared reality.
Now, the misinformation multiplier, this is where large language models, or LLMs, this is where they can generate highly convincing false information. And that often exploits cognitive biases and the ideological leanings.
Now, research shows that the AI generated disinformation is actually harder to detect than human written misinformation.
And there's the whole hallucination problem.
AI confidently invents, quote unquote facts.
And by facts, quote unquote, I mean fake citations, false legal precedents, imaginary statistics. And these hallucinations, well, they aren't bugs, they're features of how the technology performs.
There's the trust erosion. This is where the studies show that exposure to AI generated misinformation reduces trust in institutions and influences decision making. Even when people are told that the content might be AI generated.
And the watermark challenge. This is where the technical solutions like digital watermarking, they help. But there's sophisticated actors out there and they can usually circumvent those.
Now, in 2023, Dictionary.com made the word hallucinate their word of the year, specifically because of AI.
When the dictionary is calling out your technology for making stuff up, maybe there's a credibility issue. Right now, healthy skepticism is essential. And cross reference, well, the that's important information that needs to be verified, especially when it comes to health and financial advice with these sort of authoritative human sources.
But there's another side to this situation as well. Because beyond individual harms, AI poses systemic challenges to jobs, security and even human control.
There's job displacement where AI adoption is accelerated faster than previous technologies.
And while new jobs may emerge, the transition could be disruptive, especially for knowledge workers, cyber offenses. This is where the AI powered hacking tools are getting more sophisticated.
Recent research shows that the AI systems discovering vulnerabilities in real software, sometimes finding flaws that human researchers missed one AI company has increased its biological risk rating from medium to high after testing showed that their system could help plan bioweapon production better than human experts. In some cases, that company is OpenAI.
There's a control question.
Loss of control.
These type of scenarios remain theoretical, but experts disagree widely on the timeline. Some consider it implausible, while others see it as the modest likelihood risk warranting serious attention.
Now here's what I want you to be able to take away from this.
AI risks the real and they're escalating, but they're not insurmountable.
The same technology creating these challenges can help defend against them.
Your awareness, your skepticism, and your willingness to engage thoughtfully. That's the first line of defense.
My favorite listener in our final episode, we're bringing it all together. How AI is transforming education, health care and innovation for the better.
Because despite the risks, the potential for positive impact is extraordinary. And that's going to bring us to today's final word.
In today's final word, AI enabled scams, deepfakes and manipulation are real threats requiring personal vigilance. Misinformation spreads faster and more convincingly than ever, eroding shared truth.
Larger systemic risks to jobs, security and even human control deserve serious attention and not really panic.
But ultimately, awareness is your armor. The More you understand AI's capabilities and limitations, the better equipped you are to benefit from it safely.
And that's today's final word.
If this episode has helped you think about AI safety differently, please do me a favor. Share this with a friend. Share with a colleague.
Visit us at PublicFacingPod stream for all of our episodes. You can email me direct at hello@PublicfacingPod stream to join the conversation.
Follow and stay connected to Public Facing the podcast for more conversations about the people in issues and everyday moments shaping our world.
Thanks for spending time with us today. Until next time, for Public Facing, I'm Joe Holman. Stay informed, stay encouraged, and stay Public Facing.
[00:10:39] Speaker A: Thanks for listening to Public Facing, the podcast with Joe Holman. Visit us online at public facing pod.stream and on
[email protected] public facing pod SA.
