In this special guest feature, Adam Carrigan, Co-founder and COO of MindsDB, discusses how the democratization of machine learning makes predictive analytics accessible to companies of any size. Adam is an entrepreneur and former Management Consultant at Deloitte. After completing his studies at the Australian National University and the University of Queensland he completed his dissertation at the University of Cambridge – specializing in the use of NLP to predict equity pricing. With significant operational experience in various industries and extensive knowledge in finance, marketing and strategy, Adam co-led his newest venture MindsDB through the YCombinator accelerator and has recently announced $4 Million in seed funding.
While machine learning makes a transformative impact on the business world, it seems larger enterprises are the only ones reaping the benefits. So what about smaller and medium-sized businesses? Are they forced to remain on the sidelines of the AI revolution?
The SMB can glean actionable information out of their corporate data stores by merely embedding an AI layer into their databases. Notably, this all gets accomplished without depending on expensive technical personnel.
What prevents smaller businesses from adopting machine learning?
Adam: As a nascent technology, machine learning remains in a maturing state. Data scientists and software engineers experienced in ML are in great demand throughout the industry and thus also command a very high price. The vast majority of companies cannot budget for these technical resources to implement ML-powered predictive analytics successfully. I conducted some research a few years back about why data scientists are so expensive noting the solution is a systemic one that will take years to correct. Thus, nothing much has changed in the two years since penning this piece.
While cost often has the most significant impact on the decision to hire a data scientist, it is not the only hurdle. Data scientists can’t operate in isolation and require the infrastructure and support to make a meaningful impact. Something that requires time, experience and significant resources to get right. It is hard enough to find a good data scientist let alone one with expertise in a given industry. So even if an SMB has the budget and can recruit a data scientist, they are almost certainly not going to understand your industry, your data and your problems. Thus, this support becomes even more meaningful to success.
How does the democratization of machine learning make a difference?
Adam: As opposed to machine learning replacing employees, the technology actually enhances the human decision-making process. By giving business analysts enhanced capabilities to derive insights from corporate data, smaller companies can make more effective choices no matter the scenario. This approach is already making a difference in many industries from financial services to retail, helping existing employees tackle problems that would usually take a data science team weeks to solve.
A new concept, AI-Tables, illustrates the power of machine learning when in the grasp of the professionals actually using data. A user can run a basic SQL query that returns predictions in a result set. In effect, the machine learning model gets queried in a similar manner as a database table.
Needless to stay, this approach lowers the expertise level needed for working with ML models. Actionable insights from a company’s data store are available to the people that need it, without requiring a data scientist or software engineer.
Can machine learning models be trusted?
Adam: The more important question is “When can machine learning models be trusted?” and equally important “When can’t they?” If the model’s creator or even the management they report to cannot determine the level of trust of the models they are using, this ultimately diminishes their power for change. People will be less likely to rely on it for decision making and thus dull its impact. For this reason, machine learning models need to leverage a concept known as Explainable AI (XAI) that details why the model returned a particular prediction. Armed with the knowledge of the underlying data, the end-user gains the ability to improve the model’s accuracy and gains insight into when the model can be trusted and when it cannot.
Over time, as more people inside the organization use AI-Tables, those underlying models become more trusted. After all, the end goal of democratizing machine learning is making data more accessible and actionable. Smarter decision-making for the business becomes the ultimate result.
AI-Tables and Explainable AI are new innovations at the forefront of making machine learning more accessible.
Sign up for the free insideBIGDATA newsletter.
Join us on Twitter: @InsideBigData1 – https://twitter.com/InsideBigData1