Since the invention of generative pre-trained transformers in 2018 (later popularized by OpenAI's marketing genius with the release of ChatGPT in 2022), the AI capabilities have been largely stagnant.
The market resolves YES if there is another significant development (likely a new technology or a discontinuous shift in GPT capabilities) by the end of 2028.
Update 2025-08-06 (PST) (AI summary of creator comment): The creator has provided additional details on what constitutes a discontinuous shift in GPT capabilities or significant development:
A discontinuous shift would be an ability to do something beyond autocomplete/stochastic autocomplete
Current models produce results that get exponentially weaker the farther away from the initial input - if this changed, it would count as significant
If models become capable of autocompleting with something creative (rather than just averaging from training data), this would be significant
@Fynn so it's all basically a stochastic autocomplete .. even the world whatever models are just autocomplete in a high number of dimensions .. discontinuous shift would be an ability to do something beyond autocomplete
there are many ways this could be done
currently, if you give a model (any model in existence) some input, the results gets exponentially weaker the farther away from the initial input the model has gone .. if that changed, I would count it as a significant development
the autocomplete only works by putting in some average of what is has seen in the training set .. if it becomes capable of autocompleting with something creative, that would also be significant