Silicon Valley overshadows its transatlantic rivals. But as artificial intelligence grows – and with a global summit on it this week – some think it could offer a Euro startup the chance to become a new Google
Europe arguably already has created very successful AI powered products. It’s not Apple level yet, but there is no denying the success.
Already quite old, but still better than most, if not all, alternatives is deepl.com to translate text. I very much prefer them over google translate, their results are just plain better.
LAION, which you may have heard of already due to their LAION-5B dataset is an organization that provides a set of images together with their captions to be used as training data. Googles imagen was trained on that for example.
It’s a German non profit organization, the product is freely available for everyone.
Also in the realm of text to image AI is stable diffusion. Arguably the best text to image model for actual art, due to the widely available additions that allow fine-tuning of the results. Midjourney makes esthetically pleasing images and DALL-E can do text now. But there is simply no other text to image model that can produce a landscape view with exactly this or that shape. Or a portrait of exactly this person.
Stable diffusion is free and open source and provided for everyone free of charge. It was developed in Germany.
A lack of hyper capitalism makes these companies fail in the metrics usually used to determine success, but that doesn’t mean they are unsuccessful or unimportant.
Europe arguably already has created very successful AI powered products. It’s not Apple level yet, but there is no denying the success.
Already quite old, but still better than most, if not all, alternatives is deepl.com to translate text. I very much prefer them over google translate, their results are just plain better.
LAION, which you may have heard of already due to their LAION-5B dataset is an organization that provides a set of images together with their captions to be used as training data. Googles imagen was trained on that for example.
It’s a German non profit organization, the product is freely available for everyone.
Also in the realm of text to image AI is stable diffusion. Arguably the best text to image model for actual art, due to the widely available additions that allow fine-tuning of the results. Midjourney makes esthetically pleasing images and DALL-E can do text now. But there is simply no other text to image model that can produce a landscape view with exactly this or that shape. Or a portrait of exactly this person.
Stable diffusion is free and open source and provided for everyone free of charge. It was developed in Germany.
A lack of hyper capitalism makes these companies fail in the metrics usually used to determine success, but that doesn’t mean they are unsuccessful or unimportant.