AI is not as smart as it thinks it is cautions Janelle Shane in her excellent blog aiweirdness.com and in her book “You Look Like a Thing and I Love You”.
AI excels at single “simple” tasks rather than multiple complex ones. One of the examples Janelle gives is about phantom giraffes. AI trained to answer questions about pictures will make assumptions based on its history. If someone asks the question “how many giraffes are in this picture” the AI will always answer at least one – the reason is that any time it was asked that question in training there was always at least one – why would you ask the question if there were not any giraffes?
By contrast, Renovite’s AI experts carefully examine the right questions for your business needs and apply the best technical solutions to address them. For example, consider the challenges of continuous ATM availability. Our Reno-Cloud solution uses separate models for cash availability; one to predict the cash required, and a second to predict if cash will run out. This strategy allows the models to run efficiently and effectively – and it ensures that models can be tuned to what they are good at. Combining the results with our heuristics engine ensures we get the best possible results.
And we never predict any giraffes when they are not there.