At the moment, artificial intelligence is still mostly being used by the well-known tech companies, because of high development- and implementation costs. However, we can recognise the trend of AI being democratised: everyone will have access to machine learning technologies. But far from all companies have access to people that are knowledgeable in data science.
The one thing that will enable AI for everyone is cloud service. Microsoft Azure, Amazon Web Services and Google Cloud are now offering AI cloud services, so that every industry can profit from this “next big thing”, not just tech companies. This has the potential to boost the economy grandly.
However, there are not nearly enough data scientists, who are just as necessary for companies as the technologies itself. The shortage is also one of the reasons why AI isn’t as widespread yet, but if the AI-cloud offerings really become mainstream for companies, it would mean that non-experts would be in control of the systems. These people do not know how to get the most value out of the data they have, and even more worrisome, they will not know how to deal with ethical dilemmas of AI. For example, the risk of having algorithms that are biased in gender or race will be much bigger in my opinion, and once in place, unskilled people will not be able to remove the biases. This will not be beneficial for the image of AI itself, so is highly undesirable. AI requires skilled and knowledgeable data scientists.
More people need to be educated in being data scientists. In the meantime, the earlier mentioned tech companies have identified a gap in the market. Google and Amazon already started with their own AI consultancies. I’m expecting that this is just the start of a trend. AI consulting will most likely become a big industry. However, this might further increase the might of the big tech companies.
[1] https://www.technologyreview.com/lists/technologies/2018/
[2] https://singularityhub.com/2018/02/19/the-democratization-of-ai-is-putting-powerful-tools-in-the-hands-of-non-experts/#sm.000189o86r6dpdgis4d164b3z1542
Dear Niels,
I think you are touching upon a very interesting development here. I totally agree with you in that AI consulting will become an industry field by itself. While the big players will certainly intend to fill this void, I would like to propose another provider of AI consulting services.
Already, today the renown consulting firms such as McKinsey or BCG have realized their clients’ need for data driven solutions. As a result, the consulting firms have been branching out and are developing their own internal data science consulting services. BCG for instance is rapidly growing its data science arm “Gamma”. However, since consulting firms might not (yet) have sufficient data science know how, I see an opportunity in partnering up with smaller, highly specialized, AI startups. Bringing together the structured approach of a leading consulting firm and the AI niche knowhow of smaller AI-software providers, could become a driver of AI consulting. Further, such teams might be able to analyze the overall situation of clients and provide a customized software solution. In this way, traditional consulting firms might be able to capture a share of the growing AI consulting industry.
Great post Niels!
Your post nicely demonstrates the risks caused by the reverse version of Polyani’s Paradox that machines know more than they can tell us.
I completely agree with you that we need more data scientist, who understand the risks of machine learning, such as the hidden biases in the data, the lack of verifiability of the data and the difficulty of correcting the errors in the data. Besides more data scientists, however, I also believe that we need more ethical experts. Data scientist are regularly experts in relatively narrow areas, but often lack any formal training in ethical reasoning.