Latest / Diplomacy and International Relations / When AI and Algorithms Rewrite the Rules of Statecraft

Transcript

When AI and Algorithms Rewrite the Rules of Statecraft · Diplomacy and International Relations

• 0:00
  Welcome to Diplomacy and International Relations.
• 0:03
  You know the show where we focus on international relations,
• 0:05
  global politics and Danish political developments.
• 0:10
  I'm your host and today we're looking at something really
• 0:13
  incredible. Yeah, we.
• 0:14
  Are I'm excited to get into this one me.
• 0:15
  Too, we're going to explore the massive structural and
• 0:19
  technological transitions that are basically, well, reshaping
• 0:23
  global sovereignty today. Specifically, we're talking
• 0:26
  about how artificial intelligence is entirely
• 0:28
  rewriting the rules of statecraft.
• 0:30
  Right. And how nations like Denmark,
• 0:32
  for example, are racing to secure both their technological
• 0:36
  and environmental futures in the middle of all this.
• 0:38
  Exactly. But hey, before we jump into the
• 0:40
  analysis, just a quick reminder for you to follow and subscribe
• 0:43
  so you don't miss our updates. You can also support the show on
• 0:46
  Patreon, which is awesome because it gets you access to
• 0:49
  bonus content and our weekly and monthly briefings on Dhemish
• 0:52
  politics. Yeah, those briefings are really
• 0:54
  worth checking out. Definitely the link for that is
• 0:56
  right down in the description. Also, I need to add a quick
• 0:59
  disclaimer here. Our sources today touch on some
• 1:02
  highly politicized scenarios, right?
• 1:05
  Things like US, China, competition, Russia and Ukraine,
• 1:08
  the China, Taiwan situation. Right, heavy topics.
• 1:11
  Very heavy. So we want to be absolutely
• 1:13
  clear with you. We do not take sides or endorse
• 1:16
  any specific viewpoints here. Our goal is strictly to
• 1:19
  impartially convey the facts and the ideas just as they are
• 1:22
  presented in the original source material.
• 1:25
  Exactly. We're just looking at the data
• 1:26
  objectively because honestly that's the only way to
• 1:28
  understand the sheer scale of what's happening.
• 1:31
  The whole landscape of international relations is
• 1:33
  shifting from human paced diplomacy to what sources call
• 1:38
  algorithmic power. OK, let's unpack this because I
• 1:41
  think a good way to visualize it is well, imagine a high stakes
• 1:44
  global chess match, but instead of grandmasters relying on their
• 1:48
  own intuition, world leaders are getting their next moves
• 1:51
  whispered to them by these lightning fast, totally opaque
• 1:54
  supercomputers. If we connect this to the bigger
• 1:56
  picture, it fundamentally breaks traditional international
• 2:00
  relations theories. They just can't explain today's
• 2:03
  landscape anymore. Right, because classic realism
• 2:05
  assumes that sovereign states are the ones holding all the
• 2:07
  power, right? Exactly.
• 2:09
  But AI creates this new dynamic of, well, liquid power.
• 2:14
  It's transnational, it's highly decentralized, and it's largely
• 2:18
  held by private tech companies, not governments.
• 2:21
  Which is wild to think about. And constructivism is struggling
• 2:23
  too, right? Oh, completely.
• 2:25
  Constructivism relies on human norms shaping reality.
• 2:29
  But now algorithms are acting as these non human norm
• 2:33
  entrepreneurs. They're basically creating a
• 2:35
  post truth algorithmic reality by deciding what information you
• 2:39
  even see. So we're moving away from
• 2:40
  traditional public diplomacy, where human analysts synthesize
• 2:44
  intelligence over like weeks or months, and instead we're
• 2:47
  relying on predictive modeling operating in milliseconds.
• 2:49
  Right, milliseconds. But wait, if leaders are relying
• 2:52
  on algorithms to make split second decisions, what happens
• 2:54
  when those artificial advisers are just, well, wrong or biased?
• 2:59
  That is the big question. The research points to a major
• 3:03
  psychological trap called automation bias.
• 3:06
  Automation bias. OK, what's that?
• 3:07
  It's basically when decision makers are under immense
• 3:10
  pressure and they just blindly trust whatever the AI
• 3:14
  recommends. Yeah, because the AI sounds so
• 3:16
  confident. Data-driven.
• 3:18
  Well. Let me push back on that a
• 3:19
  little bit. Is the machine actually the
• 3:21
  enemy here, or is the real danger just human cognitive
• 3:25
  overload? Like, maybe leaders are just so
• 3:27
  overwhelmed by data that they're basically abdicating their
• 3:30
  responsibility to a screen. I mean, the cognitive overload
• 3:33
  is definitely real, but the danger of trusting the screen is
• 3:36
  what analysts call the perception gap, because
• 3:39
  different national AI models interpret the exact same event
• 3:42
  completely differently depending on their training data.
• 3:45
  Oh wow, OK, do we have a specific example of?
• 3:46
  That yeah, the sources highlight a scenario in the South China
• 3:50
  Sea. So imagine AUS trained AI model
• 3:53
  observing a Chinese naval drill. Because of its western training
• 3:56
  data, it might flag that drill as an aggressive blockade.
• 4:00
  But a Chinese trained AI looking at the exact same patch of ocean
• 4:04
  might view a standard US patrol as a hostile incursion because
• 4:08
  it's data priority ties domestic territorial frameworks.
• 4:11
  So they're both screaming at their commanders that the other
• 4:13
  side is attacking. That's terrifying.
• 4:15
  It is, and the studies backed this up.
• 4:18
  Research shows that US and UK models are actually 25 to 40%
• 4:23
  more prone to suggesting aggressive responses in conflict
• 4:26
  simulations compared to Chinese or Russian models.
• 4:29
  Wait, really? Up to 40%.
• 4:31
  Yeah, even smaller military AI models show a 35% variance on
• 4:36
  foreign policy tasks purely based on what they were trained
• 4:39
  on. That is staggering.
• 4:40
  And it's not just happening in military simulations, right?
• 4:42
  It's bleeding into everyday life.
• 4:44
  Algorithms are actively fighting this like silent daily war for
• 4:48
  public. Opinion.
• 4:49
  Exactly. It's completely altering
• 4:50
  cognitive security through deep fakes in these massive digital
• 4:53
  echo chambers. The weaponization of information
• 4:56
  is very real. Right, the source just talk
• 4:58
  specifically about the China Taiwan information front, how
• 5:01
  are influence operations using AI there?
• 5:04
  Well, AI driven influence operations are subtly
• 5:06
  normalizing certain narratives and dividing populations from
• 5:10
  within. It's not always obvious
• 5:12
  propaganda, and analysts estimate this could sway public
• 5:15
  opinion on Cross Strait tensions by as much as 15 to 20%. 15 to
• 5:19
  20% just from algorithmic nudging.
• 5:22
  So how are traditional diplomats even supposed to fight back
• 5:26
  against that? What's fascinating here is that
• 5:28
  diplomats are actually changing tactics.
• 5:30
  Instead of fighting the system, ministries of foreign affairs
• 5:33
  are starting to treat algorithms as audiences.
• 5:36
  Audiences like they're performing for the code.
• 5:38
  Basically, yes. They've realized they can't just
• 5:40
  hack the system, so they mingle with it.
• 5:43
  They add specific tones or trending hashtags to their
• 5:45
  communications to breach these digital castle walls, so to
• 5:50
  speak. That is wild.
• 5:51
  They're crafting foreign policy to appease the algorithmic food
• 5:54
  critics on social media platforms just to get their
• 5:57
  message out. Yeah, it's a totally new era of
• 5:59
  strategic communications. But here's the thing, If whoever
• 6:02
  controls the algorithm controls the global narrative, then
• 6:05
  whoever owns the physical servers running those algorithms
• 6:08
  holds the ultimate power, right? Absolutely.
• 6:11
  And that leads us straight into the global race for digital
• 6:13
  sovereignty, because right now, AI power is incredibly
• 6:16
  concentrated. The US hosts roughly 75% of
• 6:20
  global AI compute capacity. 75%, yeah.
• 6:24
  And China has about 15%, which leaves roughly 160 other nations
• 6:28
  almost totally dependent on foreign tech.
• 6:31
  Which poses a huge risk for the global S, right?
• 6:33
  The sources call it algorithmic colonialism.
• 6:36
  Algorithmic colonialism. It means these nations become
• 6:38
  passive recipients of technologies and rules that they
• 6:42
  had absolutely no part in shaping.
• 6:44
  So what does this all mean for the rest of the world?
• 6:46
  I mean, how are other regions responding?
• 6:48
  Well, supply chains are being radically reshaped and reshored
• 6:51
  to secure access to microchips. China is pushing for total
• 6:55
  self-reliance and Europe is fighting back hard with the EUAI
• 6:59
  Act and a €200 billion AI Continent Action plan.
• 7:03
  Right. And we're seeing startups
• 7:05
  stepping up too, like France's Mistral AI.
• 7:07
  Exactly. We're also seeing the rise of
• 7:09
  the sovereign cloud, Things like Cap Gemini and Oranges Blue in
• 7:13
  France, or NATO's air gapped Google Cloud.
• 7:16
  The goal is to keep sensitive data strictly within national
• 7:19
  borders. OK.
• 7:20
  So this pursuit of sovereign control and like systemic
• 7:24
  resilience, it's clearly not just about digital assets, which
• 7:27
  actually brings us to Denmark and how they are navigating
• 7:30
  their own massive structural transformation.
• 7:32
  Right. It's a perfect parallel because
• 7:34
  achieving digital sovereignty requires the exact same
• 7:37
  foresight as achieving environmental sovereignty, and
• 7:40
  Denmark is pushing hard for climate neutrality by 2045.
• 7:43
  Yeah, and the numbers are pretty impressive.
• 7:45
  They're on track for their 70% reduction target by 20-30.
• 7:48
  In fact, EUETS emissions for Danish industry already fell
• 7:52
  15.5% just from 2022 to 2023. Which is fantastic, fantastic
• 7:57
  progress. But hitting that 2045 goal is
• 7:59
  going to require immense immediate action.
• 8:02
  It requires massive domestic infrastructure.
• 8:04
  Both green energy and bio solutions are existential
• 8:07
  requirements for national competitiveness now.
• 8:09
  Right, Things like power to X and carbon capture, utilization
• 8:13
  and storage or CCUS, plus building out cross-border
• 8:17
  hydrogen networks. Exactly.
• 8:19
  And just like the AI arms race, the labor and resource
• 8:21
  requirements are staggering. The Danish industrial sector
• 8:25
  alone will need 15,000 additional full time employees
• 8:28
  every single year up to 2035. Wait 15,000 a year just for the
• 8:33
  industrial sector? Just for that sector, yes.
• 8:35
  And on top of that, Denmark is currently only 4% circular in
• 8:38
  terms of resource use, so they need a massive boost in resource
• 8:41
  efficiency. Which explains why they recently
• 8:43
  passed the Green Tripartite Agreement, right?
• 8:46
  To map out the future of agricultural land use and make
• 8:49
  sure they can actually sustain this transition.
• 8:51
  Exactly. It's all about heavy R&D
• 8:53
  investment and cross sectoral supply chain control.
• 8:56
  OK, so whether we're talking about managing a national
• 8:58
  climate transition or navigating an algorithmic conflict, how do
• 9:02
  human institutions actually govern all of this?
• 9:04
  Like how do we manage this hyper complex fog of war?
• 9:08
  This raises an important question.
• 9:10
  The consensus right now is the urgent need for a digital Geneva
• 9:15
  Convention, likely spearheaded by bodies like the UN and NATO.
• 9:19
  And what would that actually look like?
• 9:20
  The sources mentioned 3 core pillars.
• 9:22
  Right, Yes. First is AI transparency, second
• 9:25
  is robust accountability frameworks.
• 9:27
  And the third, which is arguably the most critical, is a strict
• 9:31
  human in the loop policy for any high stakes or lethal decisions.
• 9:36
  Meaning a human always has to make the final call.
• 9:38
  The AI can't just act on its own.
• 9:40
  Exactly. But it's also important to
• 9:42
  remember that AI isn't just a threat, it can be a really vital
• 9:46
  diplomatic tool if we use it right.
• 9:48
  Oh, definitely. There's that fascinating Oxford
• 9:50
  University study on the Ukraine crisis.
• 9:53
  They utilized Twitter data from 28 different international
• 9:56
  accounts to cut through the diplomatic fog of war.
• 9:58
  Right. And they used AI to run 3
• 10:00
  specific types of analytics on that data.
• 10:03
  Descriptive which asks what is happening, predictive which asks
• 10:06
  what will happen, and prescriptive which asks what
• 10:09
  should be done. It just shows that AI with the
• 10:12
  right governance can map out these insanely complex
• 10:16
  negotiations and actually serve global peace instead of just
• 10:20
  fueling accidental conflicts. Absolutely.
• 10:22
  It's a tool. It all depends on how we choose
• 10:25
  to govern it. Which leaves us with a lot to
• 10:27
  think about. I want to leave you the
• 10:29
  listener, with one final thought to Mull over As these AI models
• 10:34
  become more advanced and agentic, are we approaching a
• 10:37
  future where diplomatic treaties are negotiated entirely by
• 10:40
  algorithms in milliseconds? While human leaders just stand
• 10:43
  by and sign the paperwork. Exactly.
• 10:45
  I mean, if that happens, who is really doing the negotiating
• 10:48
  them? Yeah, is the $1,000,000
• 10:49
  question. It really is.
• 10:51
  Well, thank you so much for your time today and thank you all for
• 10:54
  joining us for this analysis. Please remember to follow the
• 10:56
  podcast. And if you enjoyed today's
• 10:58
  conversation, we'd love it if you could leave a five star
• 11:01
  rating to help expand the show's reach.
• 11:03
  Thanks for listening everyone. We'll catch you next time on
• 11:06
  diplomacy and international relations.