Hello, everyone at Damoa! 😊
It's time to break down difficult IT news into easy and fun comics, 32nd week.
This week, the surprising news that artificial intelligence (AI) solved a mathematical problem that even genius mathematicians couldn't solve for over 10 years was a hot topic. Is it really true? From a new AI testing method to an AI price war, let's take a closer look at the comics below? 👇
(Scroll through the images.)
1️⃣ Genius couldn't solve 10 math problems for 10 years, did AI 'Astra' solve them?
OpenAI announced that its next-generation AI 'Astra,' which is still unreleased, solved 10 unsolved math and computer science problems that have been unsolved for over 10 years. They presented a 249-page proof and even verified that there were no errors with a computer. However, it has not yet undergone formal review (peer review) by the academic community, so we need to be cautious. OpenAI has been criticized in the past for exaggerating its claims.
2️⃣ "Drawing a bicycle riding pelican is over" — Kapasi's new AI test
So far, when measuring AI capabilities, they've used fun tests like "Draw a pelican riding a bicycle." Now that most AIs can pass this test easily, AI expert Andre Kapasi has proposed a new test. He gives only the first paragraph of the novel 'The Lord of the Rings' and asks to "Turn this into a 3D game world." The AI wrote 5,500 lines of code in 2 hours and actually created a 3D village. It's no longer an era of scoring tests but of judging by 'feeling.'
3️⃣ "This price for this performance?" Deephike 'V4-Flash' officially released
Deephike of China officially released its new AI model 'V4-Flash.' It significantly improved coding and automation (agent) capabilities, while the price is unbelievably low. The feature is that it only improved performance through 'retraining' without changing the design. As Chinese open models continue this low-price offensive, Western expensive AIs are getting nervous.
4️⃣ OpenAI 'GPT-5.6' price reduction — Opening act of AI price war (technology·economy)
OpenAI has significantly reduced the price of some GPT-5.6 models. This reflects the internal efficiency improvements in the pricing. With Deephike and Google joining the low-price competition, the AI industry is now entering a stage where "who can offer it cheaper" determines the winner. It's good news for users, but AI companies are facing the challenge of profitability.
I hope you enjoyed this week as well. See you next week with more useful news! 🙌
