25 Jul Why work? When AI can work for you!
From the invention of computers to now, their capabilities have soared beyond expectations. The expectation that a computer could ever have the intelligence of a human being was just a dream, but with Artificial Intelligence (AI) studies growing larger by the second, the dream is becoming a reality.
Explainer: Artificial Intelligence (AI)
The father of AI, John McCarthy, says that it is “The science and engineering of making intelligent machines, especially intelligent computer programs.” Essentially, it is the study of making a computer or software think intelligently, like the way intelligent humans think. It is done by studying the human brain and the way we learn, decide and work when faced with a problem and using the results as the basis of intelligent software and systems development.
Microsoft announced during the Build 2017 Conference that three elements are coming together to make AI an everyday modern reality:
- The computational power offered by cloud computing
- The recent machine learning and algorithm developments
- The vast amounts of data widely available
Microsoft Cognitive Services is tapping into all three of these elements to bring together a set of tools developers can use to incorporate AI functions directly into their own applications.
MICROSOFT COGNITIVE SERVICES
Over two dozen services, associated APIs and software developer kits (SDK) are offered by Microsoft’s Cognitive Services and are organised into six categories. Although some of the APIs in these categories are still in preview states, it’s not long before they become fully functional and are replaced by the current experimental APIs. The categories and some example APIs are:
Vision Services:
Image processing algorithms to facilitate personalised apps through facial, emotional and object recognition. The Face API has functions to detect human faces and compare similar ones, organised them into groups and identify previously tagged people in images. This will soon link to the Emotion API (currently in preview mode) that takes an analysed face and detects various feelings.
Speech Services:
Services that process spoken language in your applications. The Translator Speech API enables real-time speech translations within your application which will soon be enhanced by the Custom Speech Service, which overcomes recognition barrier such as vocabulary, background noise and speaking style.
Language Services:
Functions to process natural language and learn to recognize what different users want. The Bing Spell Check API is service to correct spelling errors, recognize differences (names, brands, slang etc.) and understand homophones while users type. An extension of this still under preview mode is The Language Understanding Intelligent Service which understands languages contextually. This allows your app to communicate with people in the way they speak.
Knowledge Services:
Ability to map complicated information and data with the intention to solve tasks, make intelligent recommendations and semantic search. Currently, all of these services are in preview mode but one that has been grabbing peoples attention is the Recommendations API which predicts what a customer wants based on what they have previously bought and send recommendations of what they may like and want to purchase in the future.
Search Services:
APIs based on search functionalities so make apps, web pages and other user experiences more engaging and smarter. One service here is the Bing Autosuggest API that predicts what you are going to type before you finish it. Along with this is the Bing Entity Search API, still in preview mode, that identifies the most relevant entity based on what you searched for and delivers it to you. For example, you could type in ‘restaurants near me’ and you would receive a list of all the restaurants near your current location.
Labs Services:
Services still under development and in experimental phases. One example currently underway is Project Prague, which is an SDK that will help developers incorporate gesture-based controls in their applications. You can quickly define and implement customised hand gestures to enable a more natural user experience.
USAGE AND APPLICATIONS
Most cloud services are aimed to replicate existing enterprise IT functionalities with the aim to save companies money, time and stress. In contrast to this, Microsoft’s Cognitive Services aim to bring a whole new level of functionality to applications. The way every organisation uses these AI services is completely dependent on what they want to deliver to customers and in-turn brings a new meaning to the term flexibility. One place such as Facebook would use facial recognition to detect user’s friends and enable sharing, however, another place such as the government would want to use facial recognition to catch criminals and lower crime rates. One service provides nearly unlimited functionalities, below is a graph explaining how different companies use AI services.
The IT industry uses AI in their functionalities the most, however clearly other industries are jumping on board and are starting to incorporate it into their businesses. Soon, AI will be an everyday reality for everyone and everything.
PRICING AND AVAILABILITY
Like all Microsoft Cloud Services, the Cognitive Services are billed based on what you use during that month. There are no upfront or termination fees and your pricing is based on your location. The way your usage is calculated is through how many API calls there are per service per month. You receive a certain amount of transactions a month for free and if you exceed that amount you pay for whatever else you used.
For example, one of the services mentioned above was the Face API which has three tiers of pricing. The first tier is the Free Face API, which allows up to 20 transactions per minute and a total of 30 000 free transactions per month. The second tier is the Standard Face API which allows up to 10 transactions per second, all of which are billed. The first 1 000 000 transaction cost about $1.911 per 1000 transactions. The next 4 000 000 transactions cost $1.402 per 1000 transactions. Microsoft drops the price the more you use to stay competitive.
Currently, some of the API services are unavailable in some regions, which causes problems for some companies. However, these are still new services and most are still in development. Once they are released as full functioning APIs, they will be available globally much like the other Azure services. The Face API is currently available primarily in US regions but can be accessed in West Europe and Southeast Asia.
AI is the future of not only the IT industry, but everything. Soon, we will all be using machines and applications that have an equal intelligence to their users. Microsoft is taking the lead to make AI an everyday reality.