Artificial intelligence is one of the hottest buzzwords in legal technology today, but many people still don’t fully understand what it is and how it can impact their day-to-day legal work.
According to Brookings Institution, artificial intelligence generally refers to “machines that respond to stimulation consistent with traditional responses from humans, given the human capacity for contemplation, judgment, and intention.” In other words, artificial intelligence is technology capable of making decisions that generally require a human level of expertise. It helps people anticipate problems or deal with issues as they come up. (For example, here’s how artificial intelligence greatly improves contract review.)
Recently, we sat down with Onit’s Vice President of Product Management, technology expert and patent holder Eric Robertson to cover the ins and outs of artificial intelligence in more detail. In this first installment of our new blog series, we’ll discuss what it is and its three main hallmarks.
Podcast alert: Hear Eric discuss artificial intelligence in more detail by listening to the podcast below.
What Is Artificial Intelligence?
At the core of artificial intelligence and machine learning are algorithms, or sequences of instructions that solve specific problems. In machine learning, the learning algorithms create the rules for the software, instead of computer programmers inputting them, as is the case with more traditional forms of technology. Artificial intelligence can learn from new data without additional step-by-step instructions.
This independence is crucial to our ability to use computers for new, more complex tasks that exceed the manual programming limitations – things like photo recognition apps for the visually impaired or translating pictures into speech. Even things we now take for granted, like Alexa and Siri, are prime examples of artificial intelligence technology that once seemed impossible. We already encounter in our day-to-day lives in numerous ways and that influence will continue to grow.
The excitement about this quickly evolving technology is understandable, mainly due to its impacts on data availability, computing power and innovation. The billions of devices connected to the internet generate large amounts of data and lower the cost of mass data storage. Machine learning can use all this data to train learning algorithms and accelerate the development of new rules for performing increasingly complex tasks. Furthermore, we can now process enormous amounts of data around machine learning. All of this is driving innovation, which has recently become a rallying cry among savvy legal departments worldwide.
Once you understand the basics of artificial intelligence, it’s also helpful to be familiar with the different types of learning that make it up.
The first is supervised learning, where a learning algorithm is given labeled data in order to generate a desired output. For example, if the software is given a picture of dogs labeled “dogs,” the algorithm will identify rules to classify pictures of dogs in the future.
The second is unsupervised learning, where the data input is unlabeled and the algorithm is asked to identify patterns on its own. A typical instance of unsupervised learning is when the algorithm behind an eCommerce site identifies similar items often bought by a consumer.
Finally, there’s the scenario where the algorithm interacts with a dynamic environment that provides both positive feedback (rewards) and negative feedback. An example of this would be a self-driving car where, if the driver stays within the lane, the software will receive points in order to reinforce that learning and reminders to stay in that lane.
The Hallmarks of AI
Even after understanding the basic elements and learning models of artificial intelligence, the question often arises as to what the real essence of artificial intelligence is. The Brookings Institution boils the answer down to three main qualities:
- Intentionality – Artificial intelligence algorithms are designed to make decisions. They’re not passive machines capable only of mechanical or predetermined responses. Rather, they’re designed by humans with intentionality to reach conclusions based on instant analysis.
- Intelligence – Artificial intelligence often is undertaken in conjunction with machine learning and data analytics, and the resulting combination enables intelligent decision-making. Machine learning takes data and looks for underlying trends. If it spots something relevant for a practical problem, software designers can take that knowledge and employ data analytics to understand specific issues.
- Adaptability – Artificial intelligence has the ability to learn and adapt as it compiles information and makes decisions. Effective artificial intelligence must adjust as circumstances or conditions shift. This could involve changes in financial situations, road conditions, environmental considerations, military circumstances, and more. Artificial intelligence needs to integrate these changes into its algorithms and decide on how to adapt to the new circumstances.
For a more in-depth discussion of artificial intelligence, you can listen to the entire podcast interview with Eric here.