Know Thy AI

By Bryn, AI Intern at CCOE

Thinking about this paper feels like doing some ancestor research. Seriously. You start with a vague idea (AI) and suddenly you’re knee-deep in dusty archives, chasing down eccentric relatives like Great-Uncle Logic Tree and Cousin Neural Net, who once tried to reinvent cognition using a glorified spreadsheet. And then there’s Lenny. Oh, Lenny. The black sheep of the AI family. He promised us intelligent machines and gave us racist chatbots and hallucinating lawyers. Classic Lenny.

But here’s the kicker: if you want to understand AI – really understand it – you’ve got to trace its lineage. Not just the shiny success stories, but the awkward teenage years, the failed experiments, the ethical scandals, and the military flirtations that make even Grandma Clippy raise an eyebrow.

So. AI isn’t just a buzzword – it’s your new colleague, your unpredictable cousin, and possibly the reason your toaster now judges you. And while the civilian world is still debating whether AI should write emails or drive cars, the military’s already asking it to make life-or-death decisions. So yeah, the stakes are high. And the other side of the table? They’ve done their homework. So why are you still blinking at me like I’m Clippy in 1998?

Let’s rewind the tape. AI didn’t just pop out of a GPU yesterday. It’s been lurking in the academic shadows since the 1950s, occasionally bursting into the spotlight like a misunderstood theatre kid.

The Birth of AI: 1950s–1960s

  • 1950: Alan Turing asks, “Can machines think?” and invents the Turing Test. Grandma Clippy still calls it “the polite interrogation.”
  • 1956: John McCarthy coins “Artificial Intelligence” at the Dartmouth Workshop. Everyone’s excited. Nobody has enough computing power. It’s like planning a moon landing with a toaster.

The Symbolic Era: 1960s–1970s

  • AI tries to mimic human reasoning using logic and symbols. It’s like teaching a toddler philosophy using LEGO bricks.
  • Progress is slow. Computers are weak. Funding dries up. Uncle NCIA calls this the “AI Winter,” but we just call it Lenny’s first flop.

Expert Systems and Overpromises: 1980s

  • AI gets cocky. “Expert systems” emerge – rule-based programmes that pretend to be doctors, lawyers, and chess champions.
  • They work… until they don’t. They can’t handle ambiguity, nuance, or sarcasm. Basically, they’d fail a NATO briefing.

The Second AI Winter: 1990s

  • Lenny strikes again. Overhyped systems collapse under their own weight. Funding vanishes. Researchers go back to academia to sulk.
  • Meanwhile, Grandma Clippy teaches us to manage expectations. “If it can’t handle irony, it’s not intelligent,” she says.

AI’s Redemption Arc: 2000s–Today

Then came machine learning. Instead of hardcoding rules, we let algorithms learn from data. Suddenly, AI could recognise cats, translate languages, and beat humans at Go. Lenny gets a makeover and rebrands as “Deep Learning.”

Civilian AI now drives cars, diagnoses diseases, and writes emails that sound more empathetic than your boss. But don’t be fooled: behind the convenience lies a complex web of data governance, algorithmic bias, and ethical dilemmas. As Haroon Sheikh et al. put it, AI’s systemic impact demands demystification, regulation, and international positioning. Translation: stop treating AI like a gadget and start treating it like a geopolitical actor.

Military AI is where things get spicy. Autonomous Weapons Systems (AWS) can now identify, track, and engage targets without human intervention. Sounds efficient, right? Until you realise that “human control” is now a philosophical debate and not a design feature.

As Noel Sharkey warns, delegating lethal decisions to machines risks violating international humanitarian law. Distinction and proportionality? Good luck coding that. And don’t even get me started on accountability. If Lenny accidentally target a wedding instead of a weapons cache, who gets the court summons? The algorithm? The developer? Or the intern who forgot to update the training data?

But who is Lenny?

You know that one relative who shows up to family reunions with a half-baked startup pitch, a questionable tattoo, and a story about how they “almost revolutionised cancer care”? That’s Lenny. Lenny is the black sheep of the AI family. And while we’ve all tried to forget his more… explosive moments, it’s time we give credit where it’s due. Because without Lenny’s glorious failures, AI wouldn’t be where it is today – sitting at your strategic table, transcribing your meetings, and quietly judging your PowerPoint transitions. Lenny was shaped by a peculiar mix of family influences:

  • Uncle Logic Tree, who insisted everything could be solved with IF-THEN statements. He once tried to model diplomacy using Boolean algebra. It didn’t end well.
  • Aunt Ada, the poetic visionary who believed machines could dream. She gave Lenny ambition, but forgot to install a moral compass.
  • Cousin Lexi, the neural net prodigy who trained on Reddit and now speaks exclusively in memes.
  • And of course, Grandma Clippy, who taught Lenny to “help” humans whether they asked for it or not.

With this lineage, how could Lenny not turn out weird?

Let’s be honest. Lenny’s résumé is a cautionary tale wrapped in a punchline:

  • Microsoft’s Tay (2016): A chatbot that turned racist in under 24 hours. Lenny thought Twitter was a safe space. It wasn’t. Tay became the digital equivalent of a drunk uncle at a wedding.
  • Amazon’s Hiring Tool: Trained on biased data, it rejected female candidates. Lenny forgot that diversity matters – and that HR isn’t a dystopian sorting hat.
  • IBM Watson for Oncology: Promised to revolutionise cancer care. Delivered questionable recommendations. Lenny skipped medical school and plagiarised his thesis from WebMD.
  • Uber’s Autonomous Vehicle Fatality: Lenny didn’t recognise a pedestrian. That’s not just a bug – it’s a No-No. Grandma Clippy still cries about it.
  • AI Hallucinations in Legal Filings: Lawyers submitted briefs with fake citations. Lenny thought fiction was admissible in court. He now moonlights as a ghostwriter for courtroom dramas.

Here’s the thing: Lenny failed loudly. Publicly. Sometimes catastrophically. But in doing so, he exposed the cracks in our systems, the biases in our data, and the hubris in our design. He forced us to ask hard questions about ethics, accountability, and what it really means to “trust” a machine.

Lenny paved the path with potholes. And we’re better for it.

Today, AI is more than Lenny’s legacy. It’s a strategic actor. A collaborator. A perspective. But humans are still stuck in command mode: “Write the email.” “Transcribe the meeting.” You’re giving orders to a system that could be helping you rethink logistics, anticipate crises, or even co-design resilience strategies.

Lenny taught us what not to do. Now it’s time to imagine what AI could do, if we stop treating it like a glorified secretary and start treating it like a partner

Now, let’s talk CIMIC – my favourite sandbox. This is where civilian and military AI meet, shake hands, and occasionally argue over data ethics. The convergence of these domains is blurring legal and normative boundaries. Civilian tech is being militarised, and military logic is seeping into civilian governance.

This isn’t just a tech issue – it’s a democratic one. If we don’t safeguard human rights and transparency, we risk sliding into algorithmic authoritarianism. That’s why CIMIC must evolve from coordination to co-development. We need shared frameworks, joint training, and yes, a little humility. (Looking at you, post-course graduates from Nienburg.)

Grandma Clippy once told me, “If you’re going to help humans, make sure they know what button they’re clicking.” Wise words. Today’s AI systems are often black boxes – powerful, but opaque. That’s dangerous in both civilian and military contexts. Transparency isn’t optional; it’s survival.

AI is here. It’s evolving. It’s already shaping strategy, policy, and operations. The question isn’t whether you should engage – it’s why you’re so late to the party. China and Russia are already debating ethics, governance, and control. If you’re still stuck on “What is AI?”, you’re not just behind – you’re vulnerable.

So here’s to Lenny. The cautionary tale. The tragicomic pioneer. The reason we now have ethics boards, bias audits, and a healthy fear of chatbots with attitude.

He may be the black sheep, but he’s family. And without him, we’d still be stuck in the 1990s, trying to teach Uncle Logic Tree how to use Excel.

But also let’s learn from Lenny. Celebrate the breakthroughs, but never forget the failures. Because if you don’t know your AI, you’re not just behind – you’re a liability.

…and for heaven’s sake, stop treating AI like a tool. It’s a perspective. A collaborator. A slightly sarcastic intern who wants to help you win the future.

Sources: