This edition of The Download brings you a fascinating dual focus: the evolving capabilities of artificial intelligence in tackling complex challenges and humanity’s ongoing quest to explore the cosmos. We delve into how AI is being tested against human-designed puzzles, offering a unique lens through which to understand the technology’s progress and limitations. Simultaneously, we explore an ambitious new mission to our closest stellar neighbor, Alpha Centauri, highlighting how AI is playing a crucial role in charting the course for interstellar travel.

AI Models Stumble on Intelligence Tests: Can You Outsmart Them?

Puzzles and games have long been a cornerstone of artificial intelligence development. The very concept of "machine learning" gained traction in 1959 with an algorithm designed to master checkers. Classic games like chess and Go have also served as vital proving grounds for AI’s cognitive abilities.

The progress in AI’s puzzle-solving prowess is nothing short of remarkable and rapid. In late 2024, even the most advanced AI models could only decipher a meager 18% of the notoriously tricky New York Times Connections puzzles. However, by early 2025, some of these same models achieved near-perfect scores. This dramatic improvement showcases the accelerating pace of AI advancement.

Beyond simply demonstrating progress, these puzzles offer invaluable insights into AI’s inherent strengths and weaknesses. By observing where AI models falter and where human intellect still reigns supreme, we gain a clearer understanding of the technology’s current boundaries and future potential.

To put your own cognitive abilities to the test, we’ve curated seven puzzles that have previously stumped AI models. Now, it’s your turn to see if you can outsmart the machines. Take our challenge and discover whether you possess the ingenuity to surpass artificial intelligence.

— Grace Huckins

This article is featured in our latest print magazine, which is dedicated to the world of children. Subscribe now to ensure you receive all our future issues and stay ahead of the curve in technological discourse.

AI Charts an Interstellar Course to Alpha Centauri

In a groundbreaking announcement yesterday, the Fermi Explorer Mission, a non-profit organization, revealed its intention to launch a spacecraft destined for our nearest star system by the end of 2029. This endeavor represents a monumental undertaking. Should the mission proceed as planned, the spacecraft could reach Alpha Centauri, a star system located 4.4 light-years away, in as little as 80,000 years. The trajectory for this epic journey was ingeniously devised by an AI system developed by PSI, a prominent physics research laboratory.

This innovative AI has not only identified a novel route to Alpha Centauri but also has the potential to unlock significant insights into the existence of extraterrestrial life. Explore the intricacies of how AI discovered this new path and the profound implications this mission holds for our understanding of the universe.

— Michelle Kim

MIT Technology Review Narrated: The Evolving Role of the Astronaut

Our journey into space is driven by a complex tapestry of motivations: the pursuit of geopolitical prestige, the spirit of manifest destiny, a yearning for spiritual fulfillment, insatiable scientific curiosity, and, increasingly, the allure of commercial opportunities. In the wake of missions like Artemis II, a growing body of literature suggests that these diverse justifications are coalescing under a single, unifying truth: humans possess an innate desire to explore, an inherent drive to roam.

In "The Ultraview Effect," space anthropologist Deana L. Weibel frames human space exploration as an extension of our fundamental need to embark on pilgrimages. Eiman Jahangir, a civilian astronaut, recounts a personal voyage with Blue Origin in "A Heart for Space." Meanwhile, former NASA astronaut Leroy Chiao, in "Dinner with an Astronaut," posits that humanity’s core motivation is a simple yet profound need to "know what’s on the other side." As Leroy Chiao eloquently states, "We go into space because we want to explore. We want to see what it’s like to walk on another world." However, as our presence in the cosmos expands, we are inevitably bound to carry with us the same human foibles that have historically hindered progress on Earth.

This article is now available as part of our MIT Technology Review Narrated podcast series, released weekly on Spotify and Apple Podcasts. Simply search for MIT Technology Review Narrated on either platform and follow us to receive all our new content as soon as it’s released.

The Must-Reads: A Curated Selection of Today’s Top Tech Stories

I’ve scoured the internet to bring you the most engaging, important, and thought-provoking technology stories of the day.

1. OpenAI Restricts Next Model Amid "Critical" Cyber Risk Rating: OpenAI has announced it will impose significant restrictions on its upcoming model, Astra, following internal testing that classified it as a "critical" cyber risk. The company stated this is the first model to cross its "critical" threshold. The advanced capabilities of Astra, which can automate cyberattacks, have necessitated the implementation of extra security measures. Concerns have also been raised about potential cultural issues at OpenAI that could contribute to such safety challenges.

2. Pentagon Official Sells Millions in AI Stock: Emil Michael, a Pentagon official overseeing military AI initiatives, has sold up to $25 million in Perplexity shares. This transaction has drawn criticism, with some pundits suggesting that the Pentagon is increasingly resembling a venture capital firm. This follows a previous sale of xAI stock by Michael after the Pentagon adopted Grok.

3. Data Center Backlash Costs Missouri Councilman His Seat: Voters in Missouri have ousted a councilman who championed billions in tax breaks for data center development, signaling a growing public opposition to these energy-intensive facilities. While construction spending on data centers surged in July, they are increasingly facing bipartisan public disapproval.

4. Potential Dark Matter Particle Signs Detected: Scientists may have finally observed the first signs of dark matter particles, a discovery that could prove monumental, though further evidence is still required. This potential breakthrough has significantly broadened the search for dark matter.

5. Google Nears Release of Coding-Focused AI Model: Reports indicate that Google is on the verge of releasing a new AI model, reportedly named Gemini 3.8 Flash, which is expected to rival OpenAI and Anthropic in coding capabilities. While its release is anticipated imminently, not everyone is convinced about the ultimate impact of AI on coding professions.

6. US Army Utilizes Laser to Down Drones: The US Army has successfully employed a laser weapon to shoot down three drones near the Mexico border. This advanced weapon system is capable of detecting, tracking, and neutralizing drones using directed energy. In related news, Palantir’s CEO is backing a startup led by a former Ukrainian defense minister.

7. Map Apps Differ on "Lake America" Naming Convention: Following a directive, Google and Apple Maps have implemented changes to reflect "Lake America," while MapQuest has refused. This decision has led to a significant surge in MapQuest app downloads, with the platform climbing to the top of navigation app charts.

8. AI Groups Prioritize Bioweapon Risk Mitigation: The AI industry is actively working to address the growing risks associated with bioweapons, recognizing biology as the next major frontier for AI safety concerns. Microsoft has warned that AI could potentially create "zero-day" threats in the biological realm.

9. GoPro Pivots to AI Data Centers, Stock Soars: Camera manufacturer GoPro has announced a strategic pivot into AI data centers and a merger with photonics firm Starman Optical, sending its shares up by 40%.

10. Dyson Unveils Camera-Equipped Toothbrush with Flossing Feature: Dyson has launched a new $499 electric toothbrush, the first of its kind to feature a camera integrated into the brush head, which also performs flossing.

Quote of the Day:

"If you build an entity that is vastly smarter than you, it better be on your side."

— Marius Hobbhahn, CEO of AI safety research group Apollo Research, emphasizes the critical need for companies developing superintelligent systems to maintain control over their creations.

One More Thing: Namibia’s Hydrogen Ambitions

Namibia is setting its sights on becoming the world’s first hydrogen economy, leveraging its vast untapped potential for wind and solar power to produce hydrogen and its derivatives at competitive global prices. The nation views this as a transformative strategic industry, with at least nine projects planned or under construction. One such project in southern Namibia is among the largest proposed green hydrogen investments globally.

If even a fraction of this ambitious production materializes, it could significantly boost Namibia’s economy. However, this endeavor represents a substantial gamble, as green hydrogen technology is still in its nascent stages, and the long-term demand for its products remains uncertain. The question remains: can Namibia successfully translate its hydrogen ambitions into tangible national development?

— Jonathan W. Rosen

We can still have nice things.