Killer robots. Threatening humanoids. Robo-apocalypses and evil robots taking over the world. (Just kidding.)

Movies like Blade Runner, I, Robot, and Ex Machina and sensationalized articles in the press about artificial intelligence (AI) paint AI with a scary and sinister brush. Unfortunately, AI is often summed up as something scary and negative that will steal all of our jobs, or perhaps even worse, completely and totally overpower us. 

Fact triumphs over fiction

We see from G2 data and research that the reality of AI, and its close cousin analytics, is more nuanced and rosy than the media lets on. 

As a market research analyst at G2 focused on AI & analytics I have both the pleasure of diving deep and the privilege of researching this intriguing technology. This allows me to empower others to understand the magic and mystery behind AI, offering a look into everything from the inner workings of a neural network to broad macro trends around AI & analytics.

Often, what looks like magic from the outside relies on much more “behind the curtain.” For example, some companies have come under fire since a nameless, faceless robot was looking at their personal data, such as emails, to train their everyday AI technology. 

Instead, time and time again we see how companies are employing people to manually analyze and label data. We shall discuss the ethical quandaries that ensue due to this frequent occurrence. 

“The skills gap currently limiting the adoption of AI will be eroded by developments in technology that will drive accessibility and grow potential use cases.”

Tom Pringle, VP of technology research at G2

In this piece, we’ll paint a scene, in broad strokes, of how this prediction is becoming a reality. 

The data conveyor belt

“Data is the new oil” is a common saying heard in tech today, and it’s true. 

There are a number of intermediate steps along the way to transform data into insights, including some of the following. 

• Data integration software & data preparation software

Data, whether quantitative (e.g., house prices, number of people in a given area) or qualitative (e.g., survey responses, product reviews), can’t be popped into a machine learning model like you pop a grape into your mouth. First, the data must be cleaned and systematized, ensuring that it is in the proper format and location (e.g., server, on-premises, etc.).

• Machine learning data catalogs

This allows companies to categorize, access, interpret, and collaborate around company data across multiple data sources.

• AI platforms & business intelligence platforms 

Whether you’re building your own models or using standard ones, business intelligence tools, or analytics tools, you will be leveraging some sort of platform get from data to insights.

• Insights

Once you clean the data, it’s ready to be juiced (pardon my mixed metaphor) for insights and patterns.

We are seeing exciting movement in these areas, allowing an increasingly larger group of people to work with data and unlock its full potential.

Have your data and analyze it, too

Ensuring your data is properly prepared, carefully cleaned, and immaculately integrated is not easy (or fun) work. Therefore, the rise of AI-powered tools (whether standalone like SnapLogic’s intelligent integration platform or bundled like Qlik Data Catalyst) aids users immensely. This is one example of how the process of data analysis is becoming easier and more accessible.

When we look at data management, the rise of machine learning data catalogs (data.world, IBM Watson Knowledge Catalog, and Aginity), are helping ensure that data and its subsequent analysis are reproducible and accessible. With machine learning capabilities, these data catalogs allow any end user to discover related data, which helps build a data-driven business.

AI & analytics platforms are getting smarter. The smarter the platform, the less smart users need to be. Below are a couple of meaningful developments we have seen.

• • Dataset augmentation is giving data scientists access to synthetic data and connected datasets. Smart data enrichment is helping visionary data scientists leverage data that nobody sees.

• • Smart recommendations are giving the data expert machine learning-powered recommendations for relevant insights, letting users find answers hidden deep in their data.

Talk to your data, wherever you are

Finally, we have come to the end of the data conveyor belt. 

We are living in a convenience culture, a time in which we expect things to be immediately available and accessible, data and insights included. The following are two trends to watch for how users can get insights quickly wherever they are.

• Conversational intelligence software 

Since natural language understanding has improved, we can now talk to our data, finding and exploring insights using natural, intuitive language.  With this powerful technology, users can focus on discovering patterns and finding meaning hidden in the data as opposed to memorizing SQL queries.

• Embedded

Both the manner and method of querying data have gotten upgrades. For example, analytics platforms are building integrations with commonly used collaboration tools (e.g., Oracle digital assistant on Microsoft Teams). In addition, embedded business intelligence platforms are giving software developers the tools they need to quickly include self-service analytics capabilities into business applications.

                                                       Before                                                    After

We want explainable answers and we want them now

What is the next frontier of AI & analytics?

At G2, we think the next “big thing” will be the rise of explainable answers, or the ability for users to understand why a particular answer or insight is being produced by software. Up until this point, we have been focused on the what and the how of data. In the near future, we will not only look to get answers from our data, but will also desire answers as to why our data is what it is and why specific insights or answers were generated. 

In a world where data is big and the ethical quandaries are bigger, explainability will move from being nice to have to a must-have feature. 

We have already seen movement in this direction with enterprise solutions like Tableau’s Explain Data feature, who have realized that users are looking to deeply understand their data, not just create flashy visualizations. Many startups are also focused on the problem, including Kyndi and Fiddler. Indeed, their entire business proposition is centered around the problem of explainability and are creating AI systems that can justify the reasoning behind their conclusions and results. We’ll look forward to see what the future has in store.