Illustration by Vijay Verma
Words by Jack Riewe
No matter what you thought of the last decade, you—in some way—have been affected by new technology. I saw a post somewhere recently listing out the technology that didn’t exist 10 years ago.
- The iPad
- GPS on Smartphones
- Oculus VR
- Google Chrome
It’s crazy to think about these advancements and truly mind-blowing how much these products are such a central part of our lives as we enter 2020.
With the changing of the decade, new tech will inevitably pop up and cause a ripple effect on emerging startups across the world. Not to get too meta, we want this post to focus on machine learning and data science and how this technology will be perceived in the next decade. Us at Obviously AI deeply believe data science can be effortless for everyone regardless of their technical background or industry. We envision more data-driven decisions and predictions being made to solve problems on the enterprise level and the personal level.
Next year, we’re opening up signups to everyone. Even with the free version of Obviously AI, users will be able to make predictions and analyze their data just like they conduct a Google search. We don’t predict there will be this overnight metamorphic change in the ML world, but we do predict it will be the start of a movement to make the technical ML process more accessible, inclusive, and collaborative for everyone.
In the same way Squarespace, Figma, or Sketch changed the design industry, we predict there will be a fundamental shift in the way people think about the ML process. No-code data science will put the power in the hands of the creative product manager, the analytical marketer, the bootstrapping CEO, the recent college-grad designer, and many more.
Obviously AI turned 1 this year and we have learned a lot about how our users use our product, the no-code space, and how no-code ML fits into tech democratization and the future of data science.
Taking what we learned this year, here is a list we compiled of our best predictions for the next decade of no-code data science.
1. The General Public Will Crack Down Hard on Bias Algorithms
This was a very hot topic this year and after the Apple Card scandal, programmers received tons of flak for building a bias algorithm. Once a tweet called out Apple Card for being bias, the public hammered the algorithm that gave men a bigger credit line than women. This was a great unionization of the Twitter community to call out bias, but it also heated the debate on what makes an algorithm bias. Is it the data or how the algorithm was built?
Especially in this Twitter thread. On the heel of the Apple Card debacle the New York Times decided to publish this article. A long debate followed.
Both data and algorithms can be biased. While algorithms are trained by data—which can exclude minorities and have incomplete/inaccurate variables—they are built by one or a few humans who have their own bias blindspots. To avoid making unethical business decisions, companies now have to learn the best ways to avoid bias in machine learning.
We believe the best way to do this is to:
- Employ a data governor to oversee the data collection process.
- Make the ML process collaborative and transparent by putting it in human language.
- Take out the code so the only bias you have to worry about is in the dataset.
This will be an ongoing debate in the next decade. It will be interesting to watch how business users involve no-code ML tools in deleting bias.
2. Roles and Responsibilities Will Change the More We Use No-Code Tools
Another debate this year was the roles and skillset of no-coders in future jobs. We wrote a post on the future of no-code jobs and predicted new roles will emerge like
- User Scientist
- Data Governor
- Product Hacker
These roles implement no-code technology in their stack and experience with no-code tools in job descriptions will be required.
I came across this infographic recently which describes the responsibilities of a data scientist:
With no-code data science, these skills aren’t required. We don’t mean that in an end-of-the-data-scientist kind of way, but the role will morph as no-code becomes more adopted.
3. The SQL Query Will Disappear
Once again, we’re not expecting this to happen overnight. But, with No-Code Data Science, SQL queries are simply not required.
With the deletion of the SQL query and the introduction of NLP tech, those who use machine learning or data analytics can communicate insights visually or in human language. Additionally, without the need of the SQL query or coding language, teams can be independent of data science teams allowing them to be more agile and transparent with their findings.
4. The Machine Learning Process Will Become Collaborative
By allowing the process to become collaborative, the ML transparency improves and decreases possible bias blindspots. The traditional machine learning process is only governed by a few with proper technical skills. The ML process instantly becomes collaborative with a no-code ML tool. With no-code machine learning, predictions and analytics can be done in English by any team member. Team members can simply be added to platforms the same way as a Google Sheet. Cross-team collaboration instantly improves and the ability to scale without a data-science team increases by allowing anyone to be involved in predicting and analyzing data.
5. The Way We Look At and Talk to Data Will Change
As no-code data science becomes the new norm, how the general public treats the data process will change. If we begin to look at analytics the same way we would Google for enterprise data, data insights will become accessible to everyone with a dataset. As NLP tech improves in the future, how we talk to and visually interpret data will improve tremendously.