AI isn’t just eating tasks—it’s chipping away at what we trust, how we work, and who we believe. Here’s what that means for work, fraud, and weapons, and how to respond without panic.
Loading…
Playbooks on AI and Web3 product marketing, GTM and distribution. No spam, unsubscribe anytime.
Written by Kevan Shah, Web3 and AI product marketing strategist and founder of Hangryfeed.
Head of BD at Zealy and growth operator with 8+ years experience helping AI and Web3 products raise $26M+, scale to $2.2B monthly DeFi volume, and reach 1.2M+ users.
“It’s becoming increasingly difficult to have confidence in what we see and hear.”
That line sums up where a lot of people find themselves with AI right now. It’s not just about tools making work faster or cheaper. It’s about a deeper unease: when a video, a voice, a photo, or even a message from your boss might be synthetic, what does “evidence” even mean anymore? And what does that do to jobs, to politics, to war?
This isn’t a story about abstract “AI disruption.” It’s about three very human tensions:
Let’s walk through each, then end with what you can actually do if you build products, run teams, or ship campaigns.
If you talk to people across creative fields, software, and marketing, you hear the same pattern: AI isn’t wiping out every job, but it is devouring the middle.
One recurring theme is the loss of “good enough” work.
The result is harshest for people whose value lay in being faster or cheaper at repeatable tasks. When a model can hit 70–80% quality instantly, the market doesn’t need as many humans at that tier.
At the same time, a different pattern shows up: jobs don’t vanish; they mutate.
Many people describe AI as stripping away the “grunt work” and pushing them upstream, into roles that demand judgment, context, and originality.
There’s also a quieter but important upside: lowered barriers. People with limited training, disabilities, or fewer resources can now:
AI doesn’t automatically hand out careers. But it does hand more people the ability to make things that used to be gated behind skill, time, or money.
The tension for workers is clear: if you sit in the middle—doing competent, repeatable, mid-tier work—you’re being squeezed from both sides. The machines eat the bottom. The humans who adapt climb to the top.
If you’re a founder, operator, or developer:
Design roles around uniquely human leverage.
When you scope jobs, ask: what parts of this work are about judgment, relationships, or taste? Double down there. Let AI handle the rest.
Be honest about the “good enough” layer.
Some services you sell today will be undercut by AI-generated alternatives. Plan for that, don’t pretend it won’t happen.
Invest in skill transformations, not just tools.
Give your team time and training to climb up the value ladder. AI adoption without reskilling is just disguised layoffs.
The line “it’s becoming increasingly difficult to have confidence in what we see and hear” is more than a feeling. It’s the core problem of AI impersonation.
Deepfakes and voice cloning have made it trivial to:
On one side, there’s the obvious harm: fraud and scams.
On the other side, there’s something more corrosive: the erosion of evidence.
Even when a clip is real, anyone caught on camera can now say, “that’s just AI.” The same technology that makes fake media more convincing also hands bad actors a built-in excuse. The result is a dangerous symmetry:
This is more than a technical issue; it’s a social one. Courts, journalists, and everyday people have long treated recordings as strong proof. That norm is breaking.
In daily life, you can feel the shift:
As AI impersonation improves, the old advice—“trust your eyes and ears”—stops working. We’re entering a world where defaults invert: you assume media is fake unless proven real.
If you run a business or ship products, you can’t ignore this.
Add verification layers.
For sensitive actions like payments, data access, or account changes, don’t rely on voice alone. Use multi-factor checks, known-safe channels, or shared passphrases.
Set internal norms for media evidence.
Make it explicit inside your company that videos, screenshots, and audio are starting points, not proof. Define what counts as verification.
Brand your authenticity.
If you’re publishing content—campaigns, announcements, interviews—consider consistent signatures or formats that your audience learns to expect. Over time, they become a signal of “this is really from us.”
Educate customers without fearmongering.
Simple, clear guidance like “we will never ask you to transfer money based solely on a phone call” can prevent a lot of harm.
The third rail in this conversation is AI in weapons.
AI is already used in surveillance, targeting, and battlefield logistics. That alone shifts how wars are fought. But the real anxiety comes from the move toward autonomous systems making life-or-death decisions.
The core worries are:
Loss of human judgment.
When an algorithm decides whom to target, “human in the loop” can become a checkbox rather than a real veto. Split-second decisions get delegated to models trained on incomplete data.
Escalation and accessibility.
As AI makes it easier to plan, coordinate, and execute attacks, smaller actors can wield more power. You no longer need massive infrastructure to act with outsized impact.
Blurry accountability.
If an AI-controlled system misidentifies a target, who’s responsible? The commander? The developer? The vendor? Responsibility spreads thin just as the stakes rise.
At the same time, some argue that ignoring AI in warfare is unrealistic. If adversaries use it, you can’t simply opt out. That produces a grim logic: “we need AI to defend against AI,” even if everyone agrees the whole arrangement is unstable.
Most startups won’t ship weapons. But the same decisions that make weapons dangerous—speed, scale, lack of oversight—show up in civilian AI systems too.
Don’t build for unchecked autonomy in high-stakes domains.
For anything touching health, finance, law, or safety, keep humans genuinely in the loop with authority to override.
Be explicit about forbidden uses.
Include clear, enforced restrictions in your terms about military, surveillance, or harm-enabling use. It won’t stop everyone, but it’s better than pretending neutrality.
Design for auditability.
If your system can make consequential calls, design logs, explanations, and controls so that someone can answer “why did it do that?” when things go wrong.
It’s easy to spin these themes into doom: jobs collapsing, truth dissolving, weapons getting smarter. But many of the people closest to these changes aren’t nihilistic. They’re pragmatic. They see both the damage and the leverage.
If you’re a founder, operator, developer, or marketer, here are some grounded takeaways.
AI is phenomenal at pattern-matching and repetition. It is terrible at owning outcomes.
Ask yourself and your team:
Design your work so the second category grows over time. If your business is built entirely on the first, start experimenting now—before a competitor or a model undercuts you.
In a world of synthetic media and voice clones:
Trust used to be a passive default. It’s becoming an active discipline.
Ethical constraints and safety checks are often framed as obstacles. They can also be features.
No one has firm answers to the biggest questions:
You don’t get to wait for certainty. You move with partial information, you course-correct, and you remember that not all change is progress—but not all change is catastrophe either.
The challenge in this moment is to be neither naive nor paralyzed: to see clearly how AI is reshaping work, trust, and conflict, and still choose where you can build, protect, and improve.