The world can’t get enough of gesture control. Our talk was so popular at SXSW that we decided to make it a webinar to share our knowledge - so thanks to those who joined us for “How To Incorporate Gesture Control, AI, and NLP into your own J.A.R.V.I.S.”!
You can watch the recording anytime here. We’ve also curated our favorite questions from the Q&A below.
Webinar Q&A: Security, Buy-in And More
Q: Is there a risk to exposing all of your data to an AI system?
A: There is always a risk when you are opening up your data to a third party tool. Luckily most of these systems are secure, and by identifying the datasets that are needed ahead of time you can reduce to amount of exposure to potential security risks. By simply employing best practices like SSL, HTTPS, and keeping your system software up to date, most scenarios are safe.
The bigger risk we’re seeing is around new startups and manufactures that do not have the proper experience building solutions that are secure. For example, many IoT devices use system passwords like “sysadmin” and “admin,” but don’t offer the ability to change them. This has been the root of most IoT hacks and will continue to be so until there are security best practices put into place.
Q: What’s the biggest misconception about AI?
A: Everyone thinks AI is what we see in the movies (think Terminator and I, Robot ). The thing is, AI replicating human behavior isn’t actually useful to business. Instead you want specialized AI that is tailored to your problem, whether it’s crunching data, predicting trends or organizing datasets.
Q: Gesture control is really cool, but it feelS A LITTLE FUTURISTIC AND TOO EXPENSIVE. WHAT'S THE REAL WORLD BUSINESS CASE?
A: Gesture control solutions are built on the same core technologies you've been using for years - this is just the next level and the natural progression of things. Don’t get left behind. To build a business case, you can build a POC with a virtualizer and then use a working demo to secure buy-in.
Q: What other use cases do you see for gesture control?
A: Gesture control is just another way to interact with your digital systems. You can think of them as supercharged hotkeys.
Q: Voice control doesn’t seem to be all that useful when you look at systems like Alexa. Do you really see it being widely adopted?
A: Yes. Even though voice controlled AI like Alexa and Siri are limited today, Amazon and Apple have educated the general public on the value of digital assistants. As NLP improves, more and more systems will integrate voice control and commands into day to day operations. In the very near future, all of your customer service requests will be handled by AI using NLP.
Q: Which is better with Built.io: Alexa or Google Home?
A: It depends on the end goal of your application. If a simple input structure works (i.e. what’s the largest deal in the pipeline), then Alexa will meet your needs without the additional work required to implement API.ai. If your application requires more flexible input (i.e. a less rigid input structure with multiple variables), API.ai is your best option.
Q: Why are APIs so important to the adoption of AI?
A: AI functions on the easy availability of data in massive quantities. APIs are the easiest way for computers to receive and provide this data.
Q: How do we improve AI?
A: Most would say we should improve algorithms, but I see the real issue being datasets. We have to feed AI a variety of data that ensures we don’t induce bias and that the data is diverse enough to include outlier cases.
Q: How do we know that AI is making correct decisions?
A: Testing and improving AI is critical. Use sample datasets to interpret, perfect, and analyze your AI. It's important to remember that an AI is not impartial – it works based on the dataset you provide, so any biases inherent in your data come with it.
Q: Say what? My AI is biased? How?
A: The datasets that are provided to your AI can be biased. If you take historical loan data, you will find that there was discrimination based on race or gender. Now if you take this dataset and feed it into your AI, it will reach the same conclusion as the historical data. This can be prevented by monitoring the algorithms and the outputs. It’s imperative to monitor your tools to ensure that it is not becoming biased. (Based on a real story).
Q: How do you see your platform playing a role in the future of AI?
A: We see Built.io Flow as the glue or connector for AI. You can use our platform to funnel and filter data from all of your data sources like social media, financials, databases, etc. to your AI tool. From there, we can help deliver the information from the AI tool to your computer, phone, or digital assistant.
Q: How do chatbots work on Built.io?
A: We have done a lot with chatbots recently, especially with messaging services like Cisco Spark, text messages, and voice based endpoints like the Echo or Google Home. Our platform makes it simple to connect complex systems like Salesforce, Marketo, or API.ai to your chatbot regardless of what services or endpoints you choose.