ANNEX XIII
Criteria for designation of AI general purpose models with systemic risk referred to in Article 51
In determining whether a general purpose AI model has capabilities or an impact similar to those listed in Article 51(1)(a), the Commission shall consider the following criteria:
|
a) |
The number of parameters of the model; |
|
b) |
The quality or size of the dataset, measured, for example, by tokens; |
|
c) |
The amount of computational resources used to train the model, measured in floating point operations, or expressed by a combination of other variables, such as estimated cost, time or energy consumption for training; |
|
d) |
The input and output modalities of the model, such as text-to-text (large language models), text-to-image, multimodality, and the advanced thresholds for determining high-impact capabilities for each modality, and the specific type of input and output (e.g., biological sequences); |
|
e) |
benchmarks and evaluations of the model's capabilities, including the number of tasks without additional training, its adaptability to learn new, distinct tasks, its degree of autonomy and scalability, and the tools it has access to; |
|
f) |
Whether it has a major impact on the internal market because of its reach, which is assumed when it has been made available to at least 10,000 registered business users located in the Union; |
|
g) |
The number of registered end users. |
