Ensure your AI project succeeds.

Deep learning is a tool. Like any other AI tool, it has limitations:

• Results can lack explanations that users trust or understand

• Obtaining sufficient data can be a severe bottleneck

• Results can be incorrect and sometimes bizarre

• A deep learning model may reflect biases inherent in the training data

• Brittleness: data can drift, and real-world applications often have unexpected edge or out-of-domain cases

Neurosymbolic approaches mitigate these problems by combining complementary strengths of deep learning and symbolic reasoning.

Deep learning models have made amazing progress in perceptual understanding, allowing incredible practical improvements in text processing, machine translation, speech recognition, and computer vision. Still, they do not understand the world, exhibit common sense, or even understand their own limitations.

PostModern AI handles not only deep learning projects, but also next-generation projects that require neurosymbolic approaches to better support explanation, deductive inference, planning, and meta-level reasoning.