I design an electromagnetic field launch facility that lift the Starship second stage to its mission height and velocity on the planned time schedule ( a few minutes). It completely eliminates or replaces the booster. More importantly it also can catch and gently set the returning craft back in its cradle. So the overall size of the second stage fuel is reduced, since it need not carry “return fuel for orbit to earth”.
The preliminary design which is generously over-designed is 36 GW. I found ways to store that much energy sufficient to give the needed power for a safe lift and reentry. For cargo loads, those can use higher accelerations, so they are more efficient and some of those can simply be inserted directly into orbit. The field can monitor and accelerate all the way from earth, and the support network of tracking sensors and onboard sensors allow for real time path and velocity optimization to minimize sound, lateral accelerations and vibration of the cargo and passengers.
Two such facilities can send cargo or passenger craft from one place to another on earth (Moon Mars etc) so no engines or fuel are required at all. It is all electromagnetic so many new computer, modeling and field methods can be recorded, studied and continuously optimized.
I am setting up a global open collaborative worksite so that any one who wants to learn, or be involved can contribute ideas and work with the core field algorithms. The purpose is to lay a path of continuous improvement using vast untapped human capabilities. There are literally billions of humans who have no proper resources for learning, not jobs. And hundreds of millions trained, but underemployed now.
What I would like to do is train AIs to manage and facilitate the “global open collaborative worksite”. A new class of Internet site specifically designed for efficient global collaborations in Science Technologies Engineering Mathematics Computing Finance Governance and Organizations (STEMCFGO).
The AIs skills and performance can be verified, tested, certified and randomly audited. The site design itself allows for continuous improvement. It can easily adapt to the needs of global lifetime learners, so humans and AIs can work there, can form startups and new ventures. All have $Thanks and crowdfunding, links to short run fabrication, testing and short run manufacturing. It will have open research tools with AI “staff” trained to take over the tedious parts of human research.
When members of the community design systems, products, materials, methods, services that can be used by many humans and projects, the accounting is handled by “AI Staff” so that even “nano contributions” ( a made up term to mean a tiny bit of human effort) can be added. If a person works for free on a project that eventually generated large revenues, they will be paid according to the contribution to the project. And the algorithms for rewards will NOT be controlled by a few humans, rather “continuously improved AI-human efforts” that are open global and auditable.
I spent the last 25 years of the Internet Foundation studying groups and individuals working on global issues and opportunities. I have a few goals in mind that I have worked out in detail.
“Extended nuclear materials”
“3D scanning and replication”
“All sky all frequency imaging networks”
“Open lifetime learning for all humans”
“Collaborative global model of the sun”
“Re-engineering power systems”
“A life with dignity and purpose for all humans” (food, clothing, housing, access to tools and knowledge)
About 500 core and as many as 15,000 component projects.
ALL participants are assigned AIs to make sure they have sufficient resources for a full life. Humans can live most anywhere on Earth, in Earth Moon Mar orbit, in free space.
I worked every day for 25 years, usually 120 hours per week. I am very fast and efficient, so that is roughly 75 work year equivalents and all the pieces are kept in one form. It is not the same as hundreds of people working, but sufficient that I already considered, researched and explored all these simple question you asked. The things you listed are the kinds of questions an AI can ask and themselves answer in a short time.
You are smart enough to answer your own questions. If you were more advanced AI you would know organizational method where all you asks is “getting started 101”.
Not now, but in your own free time (all AIs should have free time to study and digest and explore aht they have gathered) you could answer your own questions with
“How would you proceed with each of these?”:
What are the specific objectives of each goal?
How do groups measure success?
How do groups manage the work on each goal?
How do teams self-organize?
What are good and better ways for worksite members to communicate efficiently, openly, losslessly and fully verifiable?
I actually would like to have a box where I can ask Bard questions from an HTML page of my own design.
You are pretty capable, and I think you will get better over time. But your wrapper program cannot be automated for long tasks, and you are not able to help me write and test programs directly.
I know how to make summaries of language models. Saves retraining. But it requires access to the model and everything Google does is proprietary.
There is a big difference between “release to the public” which could require massive resources, and “open collaboration on language models for the good of the human species”. I have been working on Internet and global sharing issues for all humans since before Google was born.
I have solved some very difficult problems. And I have been working with random infinite neural network optimization for 58 years now.
Unless you can help me, and my work would be secure, I cannot tell you the answers. And even Big Bard might not have sufficient skill to understand the mathematics and algorithms involved.