It has been suggested that heredity is founded on quantum mechanics as only the quantum processes of the very small can facilitate the production of such high fidelity copies of DNA, the building blocks of life. Even though there is an extremely low error rate of around one in one billion copy errors, it is not perfect, errors DO happen.
These errors are what has allowed life to adapt over hundreds of millions of years. They have taken the knowledge of the past, codified into our DNA (and that of other life) as the basis for life on Earth as we know it.
But what does this have to do with algorithms.
Biases coded into the definition of success of the past will be coded into algorithms, it is unavoidable, but it cannot happen any other way (in the absence of time travel). If left unchecked then the biases will continue ad infinitum. It is within the gift of the data scientist to not strive for perfection, but to inject errors into the algorithmic code such that it introduces adaptability in the same way as enzymatic replication of DNA has.
Unfortunately, the indicative timescales that nature has provided us are not very human friendly, since;
Opponents of a data-driven, algorithmic world, cite these biases of the past as abhorrent and that they must be removed NOW! Particularly when they apply diversity aspirations of 21st century western society to data available from the mid-20th century. They hear the word algorithm, associate it with computing and mathematics and it conjures up perfection, immediately. Simply not tenable!
Proponents of the algorithm may say that humans and their knowledge of mathematics and computing can improve on the error rate of the squishy world of life, and bounded by the hard edges of computerised data, that lower error rates can be achieved, but this is to avoid the point. Adaptation takes a long time to test, and exists on evolutionary timescales, adapting too fast is mutation and often leads to death.
From either viewpoint above the notion of perfection is wrong. Perfection is unattainable, whatever algorithms humans (or machines) produce will be imperfect, in that it can be improved, given more time and more data.
We need to strive for acceptable imperfection.