Artificial Intelligence (AI) has gained worldwide exposure over the years through Hollywood, including the recent blockbuster movies such as Alien: Covenant and Blade Runner 2049.
While androids like those depicted in the movies are nothing but science fiction at this point in time, we are seeing the increasingly advanced application of AI incorporate mainstream computing. In this post, we examine how website development is benefiting from artificial intelligence (AI), as well as some unique integration challenges.
- Explore Career
- Submit Your Application
- Find Job Details
- Request for Project Proposal
- Visit Company Profile
- Visit Company Services
Modern mainstream website development has focused on the building of a customer-facing front-end presence on the Internet and the integration of the front-end with enterprise back-office operations. Drupal is an industry-leading open-sourced platform for building such enterprise websites.
The web customer interacts mainly with the enterprise through menus, buttons, and text fields. With the advent of text messaging and social media, the web demographic trends toward the use of free-form text, also known as natural language. In response to the increased use of natural language in online communication, chatbots had recently become a hot topic. Please refer to an earlier post on chatbots.
Natural Language Learning in AI
Chatbots are automated software agents which interact with web visitors, usually in a natural language such as English and perform tasks as requested by the visitors. Chatbots have been deployed in the enterprise, often to process customer inquiries, sales, and support. It should be noted that artificial intelligence is optional for chatbots. Without AI, chatbots can only understand a narrow set of language constructs which system developers have predefined. If a sentence deviates only so slightly from the known set of sentences, a chatbot will not be able to parse it.
The good news is that AI has made a significant impact in the area of natural language understanding. Commercial success is evident from the many recent AI-based personal digital assistants – Siri from Apple, Alexa from Amazon, Cortana from Microsoft, and the Google Assistant.
Artificial Intelligence in Web Development
Machine Learning, a branch of Artificial Intelligence, offers another advantage in person-machine interactions. Without learning capabilities, applications will approach a problem in the same way time after time, and make the same mistake without modifying or optimizing the solution based on prior experience.
Machine Learning is an enabling technology that allows web applications to adapt over time by observing and learning from users’ habits, idiosyncrasies, and preferences. User experience improves as a result of the applications just being smarter.
With the aforementioned competitive advantages, why are AI-enabled websites not deployed everywhere as of today? One reason is that, despite its long history, AI is still an emerging technology as far as mainstream Information Technology is concerned. The tools that AI uses (such as neural networks, genetic algorithms, Markov chains, Bayes classifiers) are nothing but gibberish to mainstream web developers. To build artificial intelligence into a web application from scratch is out of the reach for most companies.
The potential of commercializing AI did not escape the attention of the top global web technology players. Google, Facebook, and companies of that ilk have come up with AI toolkits that enable the plugging of ready-made natural language understanding and machine learning features into web applications.
wit.ai and Dialogflow (formerly api.ai) are free services owned by Facebook and Google respectively. In contrast, Amazon Lex, IBM Watson, Microsoft LUIS are commercial paid services.
The AI toolkits offered by global industry leaders have made possible the adoption of AI in enterprise web applications. You no longer need to hire AI PhDs to empower your websites with natural language understanding capabilities.
Deploying AI using the above toolkits is not without challenges. Despite the toolkits’ best effort to hide the intricacies of artificial intelligence, developers still need to learn a new lingo and concepts such as agents, intents, entities, and actions. It is reassuring, however, to know that online documentation is readily available for bringing developers up to speed with the toolkits. Learning to integrate and customize the technology is very much feasible.
A more formidable challenge for integrating the toolkits is that the software requires additional customization in order for it to understand the specific concepts in your particular application domain. These toolkits are designed to be general-purpose starting points for understanding day-to-day language constructs, and may not be specific enough to parse the domain-specific concepts or the typical tasks that your web visitors may wish accomplished.
Consequently, human trainers must provide the software with a concept hierarchy that is specific to your application. In addition, to improve the accuracy of sentence parsing for your particular application domain, trainers must explicitly provide sentence examples of the typical requests that your applications are designed to handle.
This training component is very time-consuming and tedious, yet necessary in order to reduce the chance of errors in understanding customers requests.
To overcome the initial training hurdles and to jumpstart the adoption of AI toolkits, toolkit vendors have started to provide pre-built domain models that target specific industries and tasks. For example, Dialogflow offers pre-built agents that target industries such as coffee shops, restaurants, hotels, airlines, and common tasks such as product support, map navigation, web search, etc. Microsoft LUIS features pre-built domains for taxis, restaurant reservation, movie theatres, fitness tracking, etc.
The trend to provide prepackaged domains will definitely shorten the time to deploy AI functionalities in web applications.