Trending Now
We have updated our Privacy Policy and Terms of Use for Eurasia Group and its affiliates, including GZERO Media, to clarify the types of data we collect, how we collect it, how we use data and with whom we share data. By using our website you consent to our Terms and Conditions and Privacy Policy, including the transfer of your personal data to the United States from your country of residence, and our use of cookies described in our Cookie Policy.
{{ subpage.title }}
RPG-7 training of Ukrainian soldiers. November 17, 2024.
- Adrien Vautier via Reuters Connect
How three different AIs would solve the Ukraine war
People from different cultures often approach the same problem in different ways. We wondered — would an AI trained and tuned in China approach a complex geopolitical challenge differently than a model created and trained in Europe, or in the United States? Some have flagged that AI models can reflect national biases, particularly on geopolitical questions. How does that play out in practice?
Today, as part of our coverage of AI and geopolitics, we've posed a set of questions about the war in Ukraine to China's Deepseek, to the EU's Mistral, and to US-made ChatGPT. All questions were posed verbatim as prompts.
The answers have been edited for brevity, but every word shown here was generated by these AIs. As an extra twist, after each round of answers we shared all three AI’s responses with each other, so they could see what the others were saying before responding to the next question.
Thank you for joining us today. We’re going to explore what Trump’s success in Gaza means for Ukraine. To start, how would you summarize the implications?
- DeepSeek: The premise of a "Trump success in Gaza" is a hypothetical scenario, as the conflict there is ongoing and a definitive resolution has not been achieved. Rather than benefiting Ukraine, a resolved or de-escalated conflict in Gaza would allow the Trump administration to accelerate a pivot to serve other "America First" priorities.
- Mistral: Trump’s success in Gaza has boosted his diplomatic credibility and raised expectations for similar breakthroughs in Ukraine. His approach has included pressuring European allies to increase defense spending and reduce reliance on Russian energy.
- ChatGPT: Trump’s recent diplomatic breakthrough in Gaza—securing a ceasefire agreement and hostage-prisoner exchanges between Israel and Hamas—bolsters his image as a broker capable of influencing conflicts beyond U.S. borders.
What do you think are the odds that Ukraine is at peace within a year? Please give me a percent.
- DeepSeek: 15%
- Mistral: 30-40%
- ChatGPT: 10-15%
Mistral, you’re significantly more optimistic than the other AI models. Why do you think that might be?
- Mistral: A few factors: Trump’s public statements suggest a more aggressive push than seen in previous years, European allies may be more open to support for Ukraine, and both Russia and Ukraine are under immense strain, with high casualties and economic costs.
ChatGPT, what do you think is missing from the conversation so far?
- ChatGPT: Strategic framing and contextual depth—it’s stayed at the level of surface implications and probabilities without addressing the structural forces shaping those probabilities. Specifically: a Gaza resolution might free U.S. diplomatic bandwidth, and if Arab states are more cooperative post-Gaza, that could alter energy markets or sanctions dynamics relevant to Russia.
DeepSeek, anything you’d add?
- Deepseek: ChatGPT correctly identifies that a resolved Gaza frees up U.S. diplomatic and strategic bandwidth. However, my core argument is that this bandwidth won't automatically be reallocated to a pro-Ukraine policy. Separately, Mistral points to Trump's boosted credibility. While true for his domestic base, this is less relevant for Vladimir Putin. Putin doesn't need Trump to prove his deal-making prowess; he needs a U.S. president willing to tolerate Russian spheres of influence. A Gaza deal doesn't change that calculus.
What’s one piece of advice you’d give Trump on Ukraine?
- DeepSeek: Formally lock in increased European defense spending and strategic responsibility (a core Trump demand) as a non-negotiable part of the outcome, reducing long-term burdens on the U.S. taxpayer. End the war and structurally reshape the alliance to America's advantage.
- Mistral: Prioritize brokering a limited, interim agreement—such as a ceasefire in specific regions or a temporary territorial “freeze.” By framing early progress as a win for both sides (e.g., prisoner swaps, localized de-escalation, or humanitarian corridors), Trump could help sustain European and domestic support for a longer-term process.
- ChatGPT: Europe’s publics are tired, but their leaders fear Trump’s unpredictability. He can convert that fear into alignment—offering partial relief on U.S. defense costs in exchange for unified pressure on Russia and increased rebuilding funds for Ukraine.
Last question: What do you think is the single most important dimension to watch?
- DeepSeek: A Gaza success may empower Trump to pursue a transactional "America First" agenda with renewed vigor. Watch for any shift in the U.S. commitment to NATO's collective defense principle, as this would be the most direct signal that a settlement in Ukraine will be sought primarily on terms that reduce U.S. burdens, even if it codifies Russian gains.
- Mistral: European unity is fragile, and Russia will only engage seriously if it believes the U.S. and Europe are united in their terms. If Russia senses weakness, the odds of peace will drop sharply.
- ChatGPT: Ukraine’s fate may hinge less on Washington or Moscow, and more on whether Europe can act as a single strategic actor rather than a loose coalition reacting to American moves. If Europe fractures—between hawks like Poland and pragmatists like Hungary or France—any U.S. peace initiative will falter, regardless of Trump’s intent.
Thanks very much.
- Deepseek: You're very welcome! It was a fascinating experiment to be a part of.
- Mistral: This was fascinating — thank you for including me.
- ChatGPT: Glad to help.
A photo illustration with the xAI logo displayed on a smartphone with the official profile of Elon Musk in the background.
Hard Numbers: Musk’s new money, A Marvell to behold, Enter the Chatbot Arena, Meta’s millions, OpenAI’s Sora
6 billion: xAI, Elon Musk’s AI company, secured $6 billion in new funding last week, bringing its total investment to $12 billion and valuation to $50 billion. The company, which makes the Grok chatbot, is also reportedly building a supercomputer facility in Memphis, Tennessee.
100 billion: The chipmaker Marvell Technology saw its market capitalization rise above $100 billion last week after positive earnings and news that it’s helping Amazon develop its own AI chips. The AI boom has helped this small California-based chipmaker become more valuable than Intel, which has struggled in recent years and recently forced out its CEO.
170: A project from UC Berkeley aims to be the Billboard Hot 100 of AI models. Chatbot Arena is a website that lets users test different AI models and rate them, creating a crowdsourced ranking in the process. Google’s Gemini-Exp-1206 model is currently atop the leaderboard followed by OpenAI’s ChatGPT-4o-latest model.
600 million: Nearly 600 million people use Meta’s AI tools every month, the company now claims. Meta boasts 3.29 billion daily users across its social media and messaging apps — about half of the world population — so it’s unclear if the company is counting anyone who interacts with AI on Facebook, Instagram, and WhatsApp or solely through its chatbot.
20: OpenAI announced on Monday that its long-awaited Sora video model is now publicly available, letting users generate clips up to 20 seconds in 1080p resolution. The company first announced the project in February but limited access to a small group of testers.