Artificial Intelligence "A La Carte"
After decades of skepticism about its real potential, today artificial intelligence (AI) is apparently entering in its Golden Age, with potentially huge socioeconomic impacts. According to a recent forecast by Accenture, AI will increase economic growth by an average of 1.7% across 16 industries by 2035, with information and communication, manufacturing, and financial services leading all industries.
Although many analysts consider AI one of the leading forces behind the fourth industrial revolution, its actual development and uptake is still limited by several barriers. First, the development of AI products involves significant R&D investments, which are not available to small or medium enterprises. Furthermore, the implementation of AI requires underlying information and communications technology infrastructures, such as access to broadband and computational power, which are not yet evenly distributed resources. Last, but not least, developing AI applications implies the mastery of advanced computer science and programming skills. As a consequence, the potential empowerment that AI can offer is currently accessible only to restricted sectors of society. Thus, providing solutions for ensuring wider access to the advantages of AI is an important challenge, which can decrease the risk that AI becomes a resource in the hands of a few players. The simplification of AI development and integration may also benefit the evolution of AI itself by enabling citizens to exert more control over this technology, enhancing their awareness of its risks, and increasing the number of AI-based applications developed for social innovation purposes.
In recent years, tech giants such as Microsoft, Google, and IBM have started to increase the accessibility of AI by offering Artificial Intelligence as a Service (AIaaS), which consists of making machine learning (such as neural networks and classifiers) and AI algorithms available to the public. Examples include Azure ML, Amazon ML, Google Cloud ML, and IBM Watson. Thanks to these efforts, an increasing number of options exist for developers to connect their own application and Internet of Things devices to ‘‘ready to use’’AI-based services.
An interesting example of this trend is the increasing availability of image recognition AI services. Google Cloud Vision and Amazon Rekognition offer a broad spectrum of solutions in this area. Other smaller companies exist that offer more specialized image recognition services, such as the AI company Kairos, which provide application programming interfaces (APIs) and software development kits for face recognition integration through the use of computer vision and machine learning. Another example is the MIT Media Lab spin-off Affectiva, which has realized a cloud-based service to analyze images and videos of humans expressing emotion. The company nViso is a further player in this area, offering AI and proprietary deep learning 3D facial technology to measure users’ emotional reactions of consumers in online and retail environments.
In addition to AI-based image recognition, other commercial services are focused on providing advanced natural language processing (NLP) and natural language understanding (NLU) capabilities to different software areas, with particular reference to bot technologies. For example, IBM’s Watson Conversation Service (WCS) is focused on automating interactions between systems and end users. By using WCS, users can define NLP aspects such as intents and entities, and simulate entire conversations. Another key player in NLP/NLU technologies is Microsoft’s Language Understanding Intelligence Service (LUIS), a component of Microsoft Cognitive Services that is focused on creating and processing natural language models. Obviously, Google could not be missed from this list. The tech giant’s ace in the game is Google Natural Language API, which offers capabilities such as intent-entity detection, sentiment analysis, content classification, and relationship graphs. A more specialized service is Recast.ai (recently acquired by SAP), which has created a tool that allows developers to build, train, deploy, and monitor intelligent bots, as well as connect them to most popular messaging channels.
However, for the ‘‘democratization’’ of AI to happen, it is also necessary to simplify the way AI-based projects are designed and built, for example by easing the tools, skills, and methods required to build, deploy, and manage AI systems at scale.
These are the aims of the new AI design platform Cortex AI, the brainchild of company CognitiveScale. The platform is based on a graphical user interface (GUI) that has been designed to improve accessibility of AI capabilities. The GUI has a honeycomb-like structure, which according to the intention of its creator, the designer Mark Rolston, should allow even developers and designers without a PhD in computer science to integrate AI in their projects by taking advantage of pre-made AI ‘‘skills’’ that can perform processes such as senti- ment evaluation or NLP. Although the Cortex AI tool is aimed for companies and not individuals, it is an indicator of the increasing awareness of the need of more accessible AI- based tools, which will hopefully inspire the development of similar projects in multiple domains.