4 questions for “Stable Diffusion” inventor Björn Ommer
Posted: Sat Feb 01, 2025 4:41 am
Björn Ommer is the inventor of the AI model "Stable Diffusion", a text-to-image generator that creates images from entered commands. Ommer, a computer scientist at LMU Munich, emphasizes the importance of using AI wisely and sees it as a tool to tackle societal problems such as climate change. He advocates education across all levels to make the most of the opportunities offered by AI. Ommer freely shares his model and emphasizes the democratization of technology. He warns against a concentration of power among large tech companies and calls for AIs to be optimized to fit on smartphones in order to maintain control over the technology.
Before his lecture for the Bavarian regional group in BdKom, I was able to ask Prof. Ommer a few exclusive questions on the topics of AI, security, stable diffusion and education.
Question: What role does the open source community play in the development of AI technologies such as stable diffusion?
Answer:
The reason we are seeing such tremendous progress is because research and development is on the shoulders of giants, the previous algorithms have been published and we now have access to a huge repository of software and open source has been central to this massive growth.
Question: Will there soon be hardly any qualitative differences between open source vs. closed software?
Answer
First of all, I would perhaps talk about the opportunities lithuania phone number data and risks, and of course closed source is always said to be a way of avoiding certain risks. My point here would be that closed source at most postpones these risks somewhat.
But we have seen in the last year that closed-source approaches have been reimplemented or leaked in a very short space of time, so that we effectively only gain a small amount of time, but still run into problems.
If I look at it the other way around, open source gives me the opportunity to have transparency, openness, and in this way to better test this technology.
I wouldn't say that you can play them off against each other and say that open source is worse than closed source or something like that. On the contrary, there are large companies like Meta , for example, that release most of their software openly, while others like OpenAI are more closed.
Before his lecture for the Bavarian regional group in BdKom, I was able to ask Prof. Ommer a few exclusive questions on the topics of AI, security, stable diffusion and education.
Question: What role does the open source community play in the development of AI technologies such as stable diffusion?
Answer:
The reason we are seeing such tremendous progress is because research and development is on the shoulders of giants, the previous algorithms have been published and we now have access to a huge repository of software and open source has been central to this massive growth.
Question: Will there soon be hardly any qualitative differences between open source vs. closed software?
Answer
First of all, I would perhaps talk about the opportunities lithuania phone number data and risks, and of course closed source is always said to be a way of avoiding certain risks. My point here would be that closed source at most postpones these risks somewhat.
But we have seen in the last year that closed-source approaches have been reimplemented or leaked in a very short space of time, so that we effectively only gain a small amount of time, but still run into problems.
If I look at it the other way around, open source gives me the opportunity to have transparency, openness, and in this way to better test this technology.
I wouldn't say that you can play them off against each other and say that open source is worse than closed source or something like that. On the contrary, there are large companies like Meta , for example, that release most of their software openly, while others like OpenAI are more closed.