Artificial intelligence (AI) software has become such a staple of science fiction that most people think of it as a futuristic technology that may or may not ever become a reality.
But the truth is, most of us are already using AI software every day.
Every time you talk to your smartphone, conduct a Web search or check a social media feed, you’re interacting with artificial intelligence. AI software plays games with us, composes music and writes movies. You’re becoming more and more likely to encounter AI every time you do some online shopping as well. In fact, Gartner predicts, “By 2020, autonomous software agents outside of human control will participate in five percent of all economic transactions.”
Somewhat more ominously, the same Gartner report forecasts, “By 2018, more than 3 million workers globally will be supervised by a ‘robo-boss,'” and “By 2018, 45 percent of the fastest-growing companies will have fewer employees than instances of smart machines.”
Those sorts of predictions have some people worried about losing their jobs. Others, including luminaries like Elon Musk and Stephen Hawking, have issued warnings about the potential for AI to become dangerous. They are part of group called the Future of Life Institute, which says that intelligent machines could pose an “existential risk” to human life and is supporting research efforts dedicated to making sure that AI remains beneficial.
So far, no AI software possesses anything near the capabilities that would be necessary to endanger human life. Instead, most AI software is dedicated to far more mundane tasks like understanding natural language, identifying objects and faces in pictures, offering shopping recommendations and finding answers to common questions. However, this technology is progressing at an extremely rapid pace. Many AI agents employ machine learning techniques to become “smarter” over time, and many technology companies both large and small are actively investing in AI research.
The list below includes 45 of the most interesting artificial intelligence projects currently underway. The first half of the list focuses on the efforts of some of the world’s largest technology companies, while the second half details some of the work being done in the open source community and at smaller vendors. This list is by no means comprehensive, but it does include much of the most noteworthy AI software known to be under development.
Big Blue was one of the early pioneers of artificial intelligence, and introduced the masses to modern AI when its Watson system took part in the television game show Jeopardy. It generally refers to its AI solutions with the term “cognitive computing,” and it sells them under the brand name “Watson.” It has dozens of different AI products and services available, and they generally fall into two categories: developer tools and premade applications that use Watson technology. The company is also sponsoring a $5 million competition that challenges startups to use AI to “tackle some of the world’s grand challenges.”
1. Watson APIs
Designed for developers, these tools allow other companies to utilize Watson cognitive computing capabilities in their own apps. It currently offers about 19 different APIs with capabilities like concept expansion, conversation, language translation, personality insights, tone analyzer, relationship extraction, speech to text, text to speech, visual recognition and analytics. They can be accessed through the IBM Watson Developer Cloud service.
In the Watson Marketplace, IBM offers applications it has built that are based on its cognitive computing technology. These include Watson Trend (a personal shopper app), Watson Analytics, Talent Insights, Analytics for Social Media and Watson for Clinical Trial Matching (for the healthcare industry).
IBM has open sourced some of its machine learning technology, including SystemML. Now an Apache Incubator Project, SystemML “aims at flexible specification of ML algorithms and automatic generation of hybrid runtime plans ranging from single node, in-memory computations to distributed computations on MapReduce or Spark.”
Known for dedicating a lot of resources to research, Google has an internal team calledGoogle Brainthat works on AI projects. Much of their work gets applied to Google’s other products, including search and the Google Now Android personal assistant. It has also released some of their team’s work as open source applications, and the group has published quite a few papers on AI.
One of Google’s open source AI projects created by the Google Brain Team, Tensor Flow is a “library for numerical computation using data flow graphs.” The website includes Python and C++ APIs that allow developers to use Google’s AI capabilities in their own apps.
5. Google Cloud Machine Learning
Google makes some of its machine learning technology available to developers through its Google Cloud platform. It uses the same services for many of its own products, including Photos image search, Google voice search, Translate and Gmail smart reply.
In 2014, Google bought a London-based AI startup called DeepMind. This group’s most visible work to date has been creating the AlphaGo system, which was “the first computer program to ever beat a professional player at the game of Go.” The team is also working on applying reinforcement learning to machine learning and applying deep learning technology to the field of healthcare.
Google doesn’t provide a lot of details about the factors that contribute to rankings on its search engine, but it has said that it uses an AI technology called RankBrain as part of the algorithm. RankBrain can guess at the meaning of search terms it has never seen before and bring up relevant results. It isn’t available for download or for sale but is the subject of much interest within the technology industry.
Like the other large technology companies, Microsoft has a sizable internal team devoted to machine learning and artificial intelligence. It has subgroups focused on algorithmic economics, deep learning, machine learning, machine teaching, natural language computing and more, and it has a long list of current projects on its website. Their innovations have also been integrated into other Microsoft products and services.
Microsoft’s personal assistant software, known as Cortana, is perhaps its most visible AI product. It has been incorporated into Windows 10 and is also available for Android and iOS. It can perform tasks like providing updates, delivering reminders and handling natural language searches.
In recent years, Microsoft has begun embracing open source, and it has released some of its AI technology on GitHub. CNTK, short for Computational Network Toolkit, is a tool that allows developers to apply distributed deep learning to their own projects. It was recently updated to enable faster performance and better scalability.
10. Distributed Machine Learning Toolkit
Another open source project, the Distributed Machine Learning Toolkit (DMLT) assists with the training of big models for machine learning applications. It includes the DMTK Framework, the Light LDA topic model algorithm and the Distributed (Multisense) Word Embedding algorithm.
11. Microsoft Cognitive Services
Microsoft also offers developers several AI APIs on a subscription basis, with free tiers available. Current APIs include Computer Vision, Emotion, Face, Video, Bing Speech, Language Understanding, Knowledge Exploration, Recommendations and more. Microsoft has also used these APIs to build several sample applications, some of which have gone viral in social media.
12. Project Malmo
In this interesting project, Microsoft researchers are introducing AI to the game Minecraft. They are working to teach the AI how to make sense of complex environments, learn from others and transfer learned skills to new problem-solving challenges. It’s currently a private beta that Microsoft plans to release under an open source license.
In March 2016, Microsoft set loose an AI chatterbot named Tay that was designed to interact and learn from people on Twitter. In less than a day, other Twitter users had taught Tay to make racist and offensive comments, and Microsoft took it offline. It re-released Tay a week later only to encounter similar problems. The company says it plans to put Tay back on Twitter “once it can make the bot safe.”
The social network has invested heavily in artificial intelligence, primarily through an internal group it calls. Facebook AI Research (FAIR). Much of this research into fields like natural language processing and computer vision gets applied directly to Facebook itself through features like face tagging and newsfeed rankings. The group has also published several papers and contributes to open source AI projects.
Purchased by Facebook in 2015, Wit.ai offers developer tools for building bots that communicate with humans. Its voice recognition technology can also be used for interacting with mobile apps, home automation, wearables or even robots.
Another heavy investor in AI, Amazon has long used machine learning on its ecommerce site to make product recommendations and predict prices. And its CEO Jeff Bezos recently said, “It’s hard to overstate how big of an impact [AI is] going to have on society over the next 20 years.” In addition, the company recently purchased an AI startup known as Orbeus.
Alexa is the technology behind Amazon’s Echo device that allows users to play music, get answers to questions, buy products and more. The company recently opened up the Alexa technology to developers who can use it to power their own apps and devices.
Amazon’s cloud computing division, Amazon Web Services, offers an AI service called Amazon Machine Learning. A free tier is available for developers who want to experiment with the service, which easily scales as necessary.
Apple’s has been somewhat more tight-lipped about its AI plans than some of the other big technology firms, but there’s no doubt it is investing in the area. It has several job postings related to machine learning, and it recently purchased AI startups Emotient and Vocal IQ.
The most visible fruit of Apple’s AI efforts is its voice assistant Siri that comes installed on iOS devices. Although it has taken some criticism, this early personal assistant set the bar for similar AI-powered assistants like Cortana and Google Now.
Intel’s AI efforts have largely focused on enabling machine learning, deep learning and AI with its chips and software. It has also acquired quite a few smaller companies focused on AI.
Intel bought Saffron in October 2015. It offers two applications: Streamline uses cognitive technology to speed the development of new products, and Advantage performs visual analytics on big data.
Yahoo CEO Marissa Mayer has spoken out about the importance of AI, and the company has developed several AI tools internally to help run its various websites.
19. Caffe on Spark
This year, Yahoo open sourced a tool called Caffe on Spark that brings together two well-known open source projects (Caffe and Spark). Essentially, the project makes it possibly to perform machine learning on large Hadoop clusters, and Yahoo uses it to help run its Flickr photo service.
Acquisitions have also helped Yahoo expand its AI capabilities. In 2013, it bough SkyPhrase, and it now offers SkyPhrase natural language capabilities to developers as an SDK. However, note that developers need to request an invite in order to get access to the technology.
Like many other tech leaders, Marc Benioff, CEO of Salesforce, is enthusiastic about AI, predicting an “AI-first world.” The company has been bolstering its AI capabilities by gobbling up smaller firms.
Bought by Salesforce in April of this year, MetaMind aims to bring deep learning and artificial intelligence to business applications. MetaMind’s standalone services are being phased out as the technology becomes integrated into Salesforce’s cloud computing offerings.
PredictionIO became part of Salesforce in February 2016. It offers open source machine learning servers that developers can use to create prediction engines very quickly. Salesforce is also integrating the technology into some of its products.
This open source machine learning framework makes it easy to add audio or image processing capabilities to an application. The website includes resources like sample applications, documentation and a wiki to help developers get up to speed on the technology very quickly.
24. Apace Mahout
The stated goal of this open source project is “to build an environment for quickly creating scalable performant machine learning applications.” It includes three key pieces: a programming environment for developers who are building AI-powered applications, premade algorithms for a variety of tools and a vector math experimentation environment called Samsara.
Short for “Brain artificial,” Braina is a commercial personal assistant app for Android or Windows PCs. It has the ability to learn from information that you tell it as well as performing simple tasks on your smartphone or computer.
Developed at the University of California Berkeley and the Berkeley Vision and Learning Center, Caffe is an open source framework for deep learning. It boasts expressive architecture, extensible code, fast performance and an active community of users and developers.
The Cycorp company offers several different semantic tools under the Cyc brand name. OpenCyc is an open source knowledge base and reasoning engine; EnterpriseCyc is a commercially supported implementation of the same technology; and ResearchCyc is a free implementation for AI researchers.
This open source tool brings commercial-grade deep learning capabilities to Java. It integrates with big data tools like Hadoop and Spark, and commercial support is available through Skymind.
Encog is an open source machine learning framework that supports artificial neural networks, support vector machines, bayesian networks, hidden markov models, genetic programming and genetic algorithms. Available for Java or C#, it’s a cross-platform tool that works well on multicore, GPU-equipped hardware.
This enterprise-focused AI company counts Capital One, Cisco, Nielsen Catalina, PayPal and Transamerica among its users. It offers tools for using machine learning capabilities with big data tools like Spark, Hadoop and R, and it has both open source and commercially supported products.
The Apache Spark large-scale data processing engine has a machine learning library called MLlib. It promises easy deployment on Hadoop with 100 times faster performance than MapReduce.
This company is dedicated to creating open source technology that uses AI to control the Internet of Things (IoT). They have released several open source natural language processing tools, and they have a crowdfunded IoT control device that looks like a very friendly robot.
Neuroph is an open source Java-based framework for developing neural network architectures. It’s designed to be used by developers who are new to AI, offering quite a bit of online documentation.
35. Numenta HTM/NuPIC
Numenta is a company developing products based on a theory called Hierarchical Temporal Memory, which offers a framework for both biological and machine intelligence. NuPic is its open source platform based on this theory which can be used for data analysis, prediction and anomaly detection.
36. Open Cog
Another open source initiative, Open Cog is dedicated to “creating beneficial artificial general intelligence (AGI), with broad capabilities at the human level and ultimately beyond.” The technology is currently in use at Hong Kong Polytechnic University, and the team is confident that they will soon have software capable of human preschool-level intelligence.
37. Oryx 2
Based on the architecture of Apache Spark and Apache Kafka, Oryx 2 is an application development framework specifically designed for real-time, large-scale machine learning. It’s an open source project created by Cloudera.
Short for “Open Neural Networks,” OpenNN is a predictive analytics library written in C++ that boasts high performance. It was developed by Artelnics, a software developer that specializes in creating data analysis software for enterprises.
This open source project offers machine learning tools for Python, with a focus on data mining and analysis. It builds on the work of several other open source projects, including NumPy, SciPy, and matplotlib.
Shogun describes itself as “a large-scale, machine-learning toolbox.” It supports a wide variety of programming languages and offers classification, regression, dimensionality reduction, clustering, metric, multi-task, structured output, online learning feature hashing, ensemble methods and optimization capabilities.
According to its website, Theano has been “powering large-scale computationally intensive scientific investigations since 2007.” It’s a Python library for working with mathematical expressions involving multi-dimensional arrays efficiently, and it is useful for some deep learning applications.
Built to run on GPUs and based on LuaJIT, Torch is an open source scientific computing framework that supports a lot of machine learning algorithms. Community members have created Torch packages for machine learning, computer vision, signal processing, parallel processing and other AI applications.
Created by the team of developers behind Siri, Viv is a new AI platform controlled by conversational input. It learns constantly from the world around it, allowing it to expand its capabilities on a daily basis. The software isn’t yet available for download, but the company behind it is currently seeking partners who are interested in integrating Viv into their own products.
Created by the machine learning group at the University of Waikato in New Zealand, WEKA enables data mining in Java applications. It includes machine learning algorithms for data pre-processing, classification, regression, clustering, association rules and visualization.
This commercial project is a knowledge engine that can answer questions on a huge variety of subjects, including math, languages, chemistry, dates, health, science, money, history and much more. Anyone can use the free version on the website above, or you can subscribe to the Pro service for around five or six dollars a month.