Google PaperBanana AI and Thoughts About AI Order World

#ai #discuss #automation #techtalks

Google has released Google PaperBananaAI, an artificial intelligence tool from its new Google Banana brand product line.

The following people from Peking University are involved in the development of this project:

After the project was published under the name PaperBanana: Automating Academic Illustration for AI Scientists, Google turned it into a product.

So what exactly does Google PaperBanana do?

I am quoting their own description as is, without any changes:

Despite rapid advances in autonomous AI scientists supported by language models, creating publication-ready illustrations remains a labor-intensive bottleneck in the research workflow. To alleviate this burden, we present PaperBanana, an intermediary framework for the automated creation of publication-ready academic illustrations. Powered by state-of-the-art VLMs and image rendering models, PaperBanana organizes dedicated agents to retrieve references, plan content and style, process images, and iteratively improve through self-review. To rigorously evaluate our framework, we introduce PaperBananaBench, featuring 292 methodology diagram test scenarios selected from NeurIPS 2025 publications, covering diverse research areas and illustration styles. Extensive experiments demonstrate that PaperBanana consistently outperforms leading baselines in terms of consistency, clarity, readability, and aesthetics. We also show that our method effectively extends to the creation of high-quality statistical graphs. PaperBanana enables the automated production of publication-ready illustrations in bulk.

References:

In summary, it is described as an artificial intelligence system designed to automatically generate publication-ready diagrams and illustrations in bulk.

https://www.youtube.com/watch?v=1MqqQn5nQf8

Thoughts on the Limitations of Artificial Intelligence Use in the Scientific Field

Up to this point, I have explained what the technology itself is and what it is used for. Now, let me explain my concerns and expectations regarding this matter...

The most important question that bothers me is:

Who can guarantee that these AI bots, which summarize research and create visuals, diagrams, and statistical reports from it today, won't do research for you tomorrow?

Couldn't those who currently replace the person doing the research by creating diagrams and visuals for these publications, use those AI agents (I don't like the word "agent," a more accurate translation would be "agency") to start conducting that research themselves tomorrow?

Another concern is what will happen if we have to spend those 3-4 hours preparing these diagrams and summaries. In the past, even summarizing these scientific studies required time and effort, with scientific research being reread repeatedly. Now, give the information to an AI agent, and it will process it, summarize it, create graphs, add visuals, and hand you the final result. This has both advantages and disadvantages.

One of the advantages it provides is that it gives us back the 3-4 hours spent preparing these graphics and visuals. In other words, we can now complete a project in 3-4 hours less than it should have been. This gives us an extra 3-4 hours to use.

The negative side is that the work done in these extra 3-4 hours can be replaced, and the person who prepared the work is no longer needed to do that work.

The Problem of Humanity's Place in the Face of Changing Technologies
So how will these 3-4 hours be used? Will they again be spent benefiting technology companies? This requires serious thought:

Will we spend these extra 3-4 hours on social media platforms like Facebook, Twitter, Instagram, YouTube, TikTok, Twitch, Kick, Reddit, and WhatsApp? Will we learn about new products to buy there? Or will we learn about trends to follow?
Will we be able to dedicate these 3-4 hours to ourselves? Or will we be overwhelmed by even more workload, thinking that work has become easier?
Will we be able to spend these same 3-4 hours learning a new skill or pursuing a hobby in our own field, or spending time with friends in physical environments? How will we access sufficient money and other resources for this? How will we generate income?
Will we participate in consumption activities as consumers, providing data and resources (money, raw materials, energy) to systems? Now that the tasks and jobs we do are becoming replaceable, how will we secure a place for ourselves in this system? Will we boast about owning robots that conduct scientific research and produce products and outputs? Will we become operators of robots that conduct scientific research? If we don't have financial access to these robots and systems (the cheapest robot starts at around $5 per month, while AI robots/agents can be obtained for around $200 per month), will we become data and energy sources for these robots? (I know it sounds similar to the Battery scene in the Matrix movie, but I can't ignore this possibility either.)
What will our position be in society when the titles we define ourselves with begin to be taken away from us by machines? We live in an era where one day we are a physics researcher, the next day a research robot operator, and another day we are on the verge of being unemployed and moving to another field. (I exaggerated the proposition so that the danger is clearly visible)
Will we build lives dependent on technology companies like OpenAI, Anthropic Claude, Perplexity, Copilot, Gemini, Deepseek, Grok, Mistral, Poe, Meta, etc., which provide us with these opportunities to do things? At this point, will we have to buy or access the products produced by these technology companies, or create our own system automations, in order to do business? Many questions similar to the above await us in the next 5-10 years.

I'm ending my writing here to avoid making it too long. When I start finding answers to these questions, I will start sharing the answers I find with you again. For now, stay healthy. May God protect you!