Artificial intelligence (AI) allows us to be more efficient with our time with the ability to find patterns in a matter of seconds rather than months or years.
In recent years, we’ve heard about AI breakthroughs weekly—sometimes daily—basis. Would you believe that it’s been around for over 70 years! AI is reading mammograms and detecting breast cancer. AI can listen to speech patterns of people diagnosed with bipolar disorder and predict impending episodes based on their speech. AI can look at tax returns and public assistance applications to find likely cases of fraud and much faster than humans. And the list goes on and on and on.
So why has AI gotten so much better in the past few years? AI in the past had insufficient data, and computers were not powerful enough to run the programs. That has recently changed.
What is Artificial Intelligence and What Can’t it Do?
Here are two common myths and truths for AI:
Myth: AI is a solution to all of my problems.
Truth: AI can help make your job easier if you use it correctly, but it is not foolproof.
Myth: AI is going to replace me or my job one day soon.
Truth: AI is not going to take over the world or replace you. AI doesn’t currently have a brain nearly as smart as yours for most tasks.
AI uses computers and data to help accomplish tasks that require analysis and decision-making. It is often used to solve complex problems, increase accuracy, perform complex computations, or make decisions or recommendations. AI uses many types of data. Sometimes that means numbers in a spreadsheet but it also can use images, speech, video feeds, animal sounds, satellite imagery, or nearly anything else that can be digitized as data.
You’ll hear terms like “deep learning,” “generative adversarial networks (GANs),” “strong AI,” and “weak AI” when AI is being explained, but to get started with AI you don’t need to understand all of these.
“Deep learning” is probably the most common term associated with AI. Deep learning starts off by looking for patterns in tiny bits of data. The program then starts looking for patterns where those bits are being combined into larger chunks, and then for even more complex patterns. For example, when AI learns how to “see” images, the most basic level of AI is looking for tiny patterns that recur in images. These are the same basic shapes the receptors in our eyes see, and our brain uses to form images. The same could be done for AI that focuses on sounds or character recognition.
In the example of handwritten character recognition, you would load thousands of handwritten numbers into an AI system and let the algorithm evaluate them. The AI program could then map where each number has overlapping edges.
If you reviewed the visual output of this analysis, you would see the basic outlines of each digit, as if we had stacked thousands of ones on top of each other and they show up darkest where there is the most overlap. The AI program should be able to take any new handwritten character and see which digit it best matches. While most of the time it will get it right, when it gets it wrong and you tell the AI that it was wrong, it should add that example to its understanding and learn from it. This is why it is often called “Machine Learning.” The machine continues learning as more data is added.
You very likely have AI in your pocket, on your desk, in your car, on your wall, and maybe even on your kitchen counter right now. If you have Google Home, Amazon Echo, Siri, or another smart assistant, then you are using AI. Here are some common uses of AI:
- When you type an email and it tries to auto-complete your message
- When Facebook identifies you in a photo
- Netflix and YouTube use AI in their recommendation engines
- Waze and Google Maps use AI to determine where traffic jams are and how long it will take to get where you are going
- Nest thermostats use AI to help figure out when to warm up or cool down your house
- Spotify not only uses AI to create personalized playlists for you, but they have AI-generated music that has been listened to millions of times!
When asking “What is Artificial Intelligence”, there are many ways to classify AI. One way is to look at the function. Some functions of AI include image recognition, speech recognition, language processing, predictive analysis, and personalization.
Image Recognition and Vision
Image recognition uses vision to classify the objects in an image. Facial recognition uses vision AI to recognize individuals or fingerprints or the gait of a person as they walk. A disaster relief agency you could use AI vision to review satellite images, social media images, or weather reports to estimate damage from tornadoes, hurricanes or floods. A museum may use vision AI to quickly sift through images to identify fossils or look for photos of a certain person.
If you use Alexa, Siri, or another voice assistant then you’re using AI for speech. If you have old written documents, AI can help you turn those words into digital documents. You can also transcribe audio using AI.
Natural Language Processing
Natural language processing goes beyond speech recognition to understand the content of the speech itself. Natural language processing AI can look at thousands of legal documents and arrange them by topic. In web development, there are modules that use AI to read a PDF and then populate the taxonomy structure automatically. Sentiment analysis of news articles or social media posts about your organization is also an example of natural language processing.
Many Google products use predictive analysis to predict what you will type next. If you’re a grantmaker, predictive analysis can help predict which grant application will be successful. Fraud detection is another common use of AI. AI looks for patterns, and when it finds anomalies from its own predictions, it can point to issues like fraud.
Netflix and Amazon provide personalized experiences, and you can do the same by using data collected by cookies. If you collect information about the user, or which pages they visit, then you can push them to related content that is most likely to interest them. Some other examples include personalized museum audio tours based on which artwork visitors like or personalized volunteer schedules based on volunteer skills and availability. NGOs can use AI to send personalized donor letters.
There are many ways to use AI to speed up your work processes, personalize engagement, customize fundraising outreach, improve your digital experience, and make better use of your data and content. For more ideas on how to use AI, you can request a recording of our most recent webinar “How to Get Started with Artificial Intelligence.”