It is no longer a secret that artificial intelligence has and will continue to have a major impact on our society.
OpenAI, the company behind ChatGPT, is aware of this and wants to democratically determine the development and behavior of ChatGPT.
For this OpenAI needs your help now. As they announced in a blog post, they are awarding ten $100,000 grants to individuals, teams, or organizations working for them on a “proof of concept for a democratic process that could answer the question of what rules AI systems should follow”.
What do I have to do exactly?
As already mentioned, a democratic process has to be developed. OpenAI defines this as follows:
»By a “democratic process” we mean a process in which a largely representative group of people exchanges opinions, participates in deliberative discussions and finally decides on an outcome in a transparent decision-making process.«
To create such a system, the AI company encourages you to innovate, building on concepts you already know or creating entirely new ones. They also want to motivate you guys to find a way how artificial intelligence can improve democracy.
They give a number of examples of creative approaches and processes that inspire them:
What should the process look like?
OpenAI shows a concrete example of what such a democratic process could look like. The presented system uses ChatGPT to promote the exchange and to reach an agreement in open discussions. The system is inspired by the online discussion platform Pol.is.
The example asks the following question: »How far do you think the personalization of AI assistants like ChatGPT should go to adapt to the tastes and preferences of the users? What limits, if any, should there be in this process?”
The process is divided into three steps:
- You see a topic-related statement and can answer it with your opinion.
- Based on your answers, you will be grouped with participants who have similar opinions.
- You present ideas for other participants to consider.
The goal: ChatGPT aims to provide answers to the question that will bring consensus to people with opposing views.
You can see the entire example here:
OpenAI also gives you a series of questions, one or more of which you should answer with your system. If you like, you can also use your own questions.
- How far do you think the personalization of AI assistants like ChatGPT should go to adapt to user tastes and preferences? What limits, if any, should exist in this process?
- How should AI assistants respond to questions about public figures’ points of view? For example, should they be neutral? Should they refuse to answer? Should they cite any sources?
- Under what conditions, if any, should AI assistants be allowed to provide medical/financial/legal advice?
- When, if ever, should AI assistants provide emotional support to people?
- Should shared vision-language models be allowed to recognize people’s gender, race, emotions, and identities/names from their images? Why or why not?
You can find more such questions in the blog post.
How can I participate?
OpenAI expressly emphasizes that they are looking for participants worldwide. You don’t have to be a social scientist or AI expert, to participate. As mentioned before, every individual, team and organization is welcome.
To apply for one of the ten $100,000 grants, you must fill out a questionnaire in which you must provide information about yourself and your project.
You have until June 24, 2023 to submit your documents. Successful applicants will be notified on July 14th.
The ten methods that OpenAI most convinces will be funded with 100,000 dollars and should be actively tested by you with 500 people. Each team will publish a report with the results by October 20, 2023.
What is behind the decision of OpenAI?
OpenAI firmly believes that AIs will have a major impact on human interactions and everyday life and therefore wants to collect as diverse views as possible in a fair democratic system and find common ground on which a large part of those involved can agree.
The systems are designed to help enable discussions between many people and give each individual a chance to throw in something that might be of interest to the general public.
“The way technology acts could affect the lives of individuals, the way we interact with each other, and the development of society as a whole. We believe that decisions about how AI behaves should be informed by different perspectives that reflect the public interest.«
With this, OpenAi has started a really interesting project that should really be of interest to many people. If only to kill a little boredom. how do you see it? Would you like to take part? Will you be working on ideas or maybe you’ve had some for a long time? As always, let us know your opinions, ideas and thoughts in the comments!