For the purposes TAID, the following requirements must be met:
1. The project must have a clearly stated, overall goal. Tasks can be generated to meet the goal along the way. Focus on the goal and let the AI create the tasks to meet it.
2. The goal must be either empirical or verifiably theoretical. "What's the best sci-fi movie?" does not qualify because it is subjective opinion.
3. The goal must be something that is inherently complex enough to have stumped humans and yet also compelling enough so that humans will have tried to solve it - so that there will be lots of data for the AI to gather. No data means no goal resolution. Examples of what not to set as goals would be: How can pigs fly? This goal-from-an-expression is literally meant to convey something that is impossible. Not only is this starting with a defeatist attitude, it is something that is not seriously contemplated by anyone, not particularly interesting in real life, and would have no empirical or even established theoretical underpinning. I.e., it's a waste of time to contemplate from a social benefit perspective.
Another type of goal that would be insufficiently complex would be: "How much money would it take to solve poverty?" While curing poverty is certainly a socially worthy goal, and AI can even help, it is not a TAID because we already know how to solve poverty with money. Experts have calculated how much that would take: $175 billion per year - according to economist Jefferey Sachs, expert on world development. While there certainly can be endless debates on the exact figure - accounting for losses, corruption, etc. - these are not the true obstacles, or it would have been done already. "This yearly amount is less than 1 percent of the combined income of the richest countries in the world, and only four times the United States' military budget for one year" says the Borgen Project: How Much Does it Cost to End Poverty?
Humanity doesn't simply fund a cure for poverty this for a variety of political, social, and priority reasons that have nothing to do with intelligence - human or artificial.
To meet these three criteria, I set a goal to:
Design a faster-than-light-speed spaceship engine for a spaceship that carries 100 people for 1 year and can accelerate up to 1G continuously and has no more energy than 10 average nuclear power stations.
This is an easily defined and suitably complex goal that smart people have been working on for some time - i.e., there's lots of data - though, as we'll see, it's a goal one that can be quickly modified as needed, such as removing the energy restriction. It is theoretically verifiable; I'll be sequestering feedback from experts on this paper wherever possible. More on that later. It is both highly complex and highly desired by many humans, including very smart ones such as Elon Musk.
I didn't start my quest with Elon Musk AI, from the characters.ai website. I submitted the question to ChatGPT (version 3.5, which is still free to use), which responded thusly:
I'm sorry, but based on our current understanding of physics, it is not possible to design a spaceship engine that can travel faster than the speed of light. According to Einstein's theory of relativity, the speed of light is an absolute speed limit in the universe, and no material object can exceed this speed.
Furthermore, the energy requirements for accelerating a spaceship to such high speeds and maintaining that acceleration over an extended period of time are currently beyond our technological capabilities. The amount of energy required to accelerate a spacecraft to even a significant fraction of the speed of light would be immense, and would likely require more energy than 10 average nuclear power stations could provide.
While it is possible to design hypothetical propulsion systems that can accelerate a spacecraft to very high speeds, such as ion engines, nuclear pulse propulsion, or antimatter engines, all of these technologies are currently limited by our ability to generate and store sufficient energy, and are unlikely to enable faster-than-light travel.
Next Page 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27
(Note: You can view every article as one long page if you sign up as an Advocate Member, or higher).
1
1
1
