If you’re reading this, you probably know about how AI is changing the world of business by leaps and bounds — efficient pipelines, optimized production, automated workflows, advanced analytics that drive pronounced results, the list goes on and on. However, the waves of impact brought on by AI don’t stop at lines of profit. Sessions at our Everyday AI conferences celebrated AI changing the world around us “for good.”
In this blog, dive into these changes “for good” in highlights of sessions from Everyday AI New York, London, and Bengaluru where we learned about some of the spectacular ways that AI makes a difference.
AI Advocacy for Neurodiversity
Even as the conversation around cognitive differences expands and stereotypes start to erode, unfortunately, the barriers to hiring neurodiverse talent remain numerous. In this session from Everyday AI Bengaluru, Usha Rengaraju, Chief of Research at Exa Protocol, shared her experience in the field of autism advocacy and shared valuable insights about autism diagnosis, workspace accommodation, and how AI fits into the picture of a more inclusive future.
Acting upon diversity initiatives, many organizations express a desire to hire neurodiverse talent. But to successfully engage a workforce that is diverse in this way, an organization should work to design a workspace that supports success for everyone.
Let’s take a look at some of the key use cases where AI helps break down the barriers to inclusivity and support autistic individuals.
Early Diagnosis Using MRI Scans
Diagnosis during the early stages of neural development is crucial for autistic children as it leads to a higher probability of independent adult years. With a cognitive difference such as autism that falls on a spectrum, diagnosis is complex. Diagnosing early is very important for the implementation of correct treatments and therapy. MRI scans play a large role in early diagnosis, but analyzing these MRI scans is a highly time-consuming and difficult process when done manually. This is where AI can assist. Models based on datasets provided by the ABIDE initiative, Autism Brain Imaging Data Exchange, help doctors analyze MRI results and accurately diagnose autism quickly.
Communication Is Key — Wearable Devices Can Help
Another way that data can be used is to address communication differences for neurodivergent people, diffusing misunderstandings and avoiding harmful workplace scenarios.
When audio data is converted into mel spectrograms, image classification algorithms can be applied to these frequency visualizations. These insights can be integrated into wearable devices like Google Glass which assist neurodivergent individuals with classifying emotion signals in real time.
Text Identification and Facial Expression Detection
Often sarcasm and other indirect language tools are employed by non-neurodivergent coworkers on a day-to-day basis in meeting notes, emails, and more. Someone with autism can have difficulty identifying the nuances and connotations of indirect communication. Identifying the social cues of facial expressions can be a challenging task for many autistic individuals. This can lead to frustration and conflict. By utilizing technology aids that take advantage of text classification algorithms and facial expression detection, these problems can be avoided. Go Emotions is one dataset provided by Google that has been used in various applications. It can classify 27 different types of emotions.
Data can also be used to coach emotional detection. Take Milo the robot as an example of how AI-powered technology can be more useful compared to traditional human classroom methods of working with neurodivergent individuals. Milo is a humanoid robot trained with advanced facial expression algorithms to work in interactive scenarios with autistic students. Robots do not face the same struggles with patience that humans do which makes them more consistent and reliable behavioral coaches for autistic individuals.
Tackling Taboo and Spreading Awareness
These are just a few of the ways that AI is helping create brighter futures for autistic folks in and out of the work environment, but there is still a lot of work to be done. Part of the challenge still remains in addressing stereotypes and spreading awareness about the possibilities that do exist for neurodivergent students and workers.
The next steps will be to democratize access to these AI use cases. Making these technologies accessible and affordable globally is a challenge that organizations should be enthusiastic to take on as diversifying talent is of benefit to everyone.
Cleaning Up the Ocean: Starting On Beaches
It goes without saying that the planet we live on is the core component of a vision for a better future, and today one of the most alarming threats to our natural environment is the pollution we create. In this session at Everyday AI London, Adrien Guenard shared with us how The Ocean Cleanup works specifically on the plastic pollution problem and how AI plays a part in creating actionable solutions for tackling this daunting situation. Let’s rewind and revisit the highlights of the talk
Looking at Beaches for Big Impact
There are many strategies that exist for cleaning up the ocean, but The Ocean Cleanup sought to find the most tangible and effective way to create a lasting impact and landed on beach cleanup for this project.
Digging Up the Data
As it turns out, there is a lot of both qualitative and quantitative data that can be collected and investigated when it comes to beach cleanup activities. Who is cleaning? When are they cleaning? What do they collect? How much do they collect? Why do they care about getting plastic off the beaches? Where is the most plastic? You get the idea.
The Ocean Cleanup ended up with a ton of data on their hands and no centralized database to house all of this valuable information, so creating the world’s largest beach cleanup database was objective number one.
Creating a Beach Cleanup Database With Dataiku
The Ocean Cleanup used Dataiku to create this database and leverages the platform to focus only on plastic pollution. The data collected has to be scrutinized by weight computations to receive accurate depictions of what is actually being collected in cleanups. Then after these classifications are made, the data on the coordinates of the collection must be tracked.
Using a Google Map API, teams determine where the plastic was picked up through descriptions of areas and other key data. The Ocean Cleanup identifies “hotspots” within ten kilometers of each other with hierarchical clustering algorithms in addition to obtaining exact metrics about the density of plastic pollution for given regions. Interpreting this data helps the organization determine where cleanup projects have the most impact. Georgian computations compute the pollution density between different clusters.
What’s Next?
Taking insights gathered from initial analysis even further, next is creating a cleanup strategy. The Ocean Cleanup has developed an algorithm that enables the prediction of plastic particle position in the ocean and by coupling that with data on coastline geography and probability metrics, generates an idea of where plastic pollution might end up on the beaches. With each model calibration, the model becomes more and more efficient at finding out where plastic will show up. From here, The Ocean Cleanup teams are able to determine where beach cleanup projects should produce optimal outcomes.
Dataiku for the Future
It’s business, yes, but as you can tell from the stories above, it’s a lot more than that too! In a panel discussion at Everyday New York, we discussed the different ways that Dataiku is committed to creating a better future for everyone. Let’s recap the conversation:
What’s Ikig.AI?
Ikig.AI is a “for good” Dataiku initiative for social, environmental, and broader humanitarian impact around the world. All-in-all, it’s about giving back to our communities.
From Dataikers’ dedicated Ikig.AI work days, over six thousand lives were positively impacted. Across a wide variety of causes, this work spanned different areas from working to clean beaches to supporting nonprofits serving underprivileged youths to tackling food inequalities and taking care of local parks. Thirty-five days were used to support 24 different nonprofit partners in a variety of fields too — which brings us to the AI “for good” initiative.
An AI Initiative for Good
In the pursuit of combating climate change and championing social justice globally with the AI “for good” initiative, Dataiku has partnered with 65 nonprofits around the world. Dataiku provides these organizations with pro bono support on data science and technology.
Our Academics Program
The Dataiku Academic Program supports educational institutions by providing Dataiku licenses to the next generation of talent, the future practitioners of Everyday AI.
The Academics Program partners with St. Catherine University, the largest privately held women’s college in the United States, and Carnegie Mellon University, known for its excellent mathematics program, to recognize outstanding students for their commendable academic achievements and accomplished work.
The Dataiku Frontrunner Awards
The Frontrunner Awards were created to celebrate the success of Dataiku customers, partners, nonprofits, academics, and individual users who are pioneering new applications of AI. From identifying at-risk students to providing additional educational support to creating the first database of beach plastic cleanup and even leveraging natural language processing to study racism in the media’s handling of the global pandemic — the stories are extraordinary.
Elevating Visions and Supporting Innovation Through Everyday AI
Whether you’re working in the mundane or on your next big moonshot project, Dataiku’s mission of Everyday AI is all about expanding AI to be a collaborative resource that everyone can leverage to turn data insights into progress for positive changes in the world.