Trusting the other driver's turn signal when one or both is in motion is a good way to get killed.
Signaling to enter or exit a roundabout communicates no useful information.
It sounds interesting, but it's totally new to me.
The problem mainly occurs in networks with recurrent connections or very deep architectures. In recurrent architectures this was solved via LSTMs with the signal gates. In very deep networks, e.g. ResNet, this was solved via residual connections, i.e. skip connections over layers. There were also other advances, such as replacing sigmoid activations with the simpler ReLU.
Transformers, which are the main architecture of modern LLMs, are highly parallel without any recurrence, i.e. at any layer you still have access to all the input tokens, whereas in an RNN you process one token at a time. To solve the potential problem due to "deepness" they also utilize skip connections.
SAP is also a “German tech company”, and I bet most of their non-Germany based employees don’t speak German either.