Header Ads

Artificial Intelligence: The New Problems of the "First Mile" and the "Last Mile"

Artificial Intelligence

According to McKinsey's research institute, there are fewer than 10,000 people worldwide who have the ability to address complex artificial intelligence problems. Naturally, they are disputed among the leading companies in the branch, namely the technological ones like Google, Amazon, Microsoft, IBM and Facebook

Unlike previous technological waves where it always existed as soon as technologies were launched or became popular, resource shortages, but in which over time, many of the needs were eventually met by more intensive training programs, it is not clear that, in this case, the problem can be solved in the same way or with the same speed.

The truth is that IA is not a new area. For many years AI has been investigated, and the advancement of knowledge has been incremental and not necessarily the result of the disruption. And yet, the area still has few trained resources - and certainly, the issue will not be solved with IA certification programs. If some parallelism exists, it is with the shortage of neurosurgeons: for many training programs that are made, it will always be an area with few resources enabled. Not only for demanding specific skills but for a cycle of training and updating that is only possible for those who work permanently in the area.
A neurosurgeon does not change his area of activity every two or three years. However, in many areas of IT, the expectation is that of frequent change of area, technology, focus. Few companies have the capacity to invest sustainably in an area with uncertain yet promising results. As models become more complex, which AI is being adopted at the core of the business of organizations, there are two key issues to address - the "first-mile" problem and the "last mile" problem. The first is to obtain data to train models - especially in light of all current regulatory issues, but also to the existence of businesses where such data are intrinsically more complex to obtain. The second is the need to retrain and continually refine the models,
It is not enough, therefore, to state what has already become a cliché: "data is the new oil." Not at all, the data is the new crude, requiring a complex refining and distribution chain until it reaches the end consumer. And, indeed, of this complex chain, much, if not almost everything, is yet to be done in most companies.
Which means, of course, an opportunity. But also the awareness that it will be easy, neither fast nor cheap - and above all, it will not be solved with forced walking initiatives. To find the knowledge we seek, we need knowledge from the start - and there are no shortcuts.

No comments:

Powered by Blogger.