Artificial Intelligence (AI) allows computers to learn from experience, adapt to new inputs, and perform human-like tasks. Most of the AI examples you’ve heard about –from chess-playing machines to self-driving cars–rely heavily on deep learning, a subfield of AI. Thanks to deep learning, computers can perform complex tasks by processing vast quantities of data and by identifying data patterns.
As we observed numerous applications of AI under different business functions and industries, we believe that AI has still undiscovered potential and will enable new emerging technologies in the next few years. The increasing popularity of AI and the market growth, which we will mention in this article, support our thoughts.
We explain how AI is evolving to become more explainable and self-learning. We also outlined application areas.
What is AI?
A layman’s description would be “a machine that can demonstrate human-like intelligence in a very specific or broad field”. Since AI systems aim to mimic human intelligence, and we can measure their effectiveness by comparing them against the best performing humans in the same field.
The Oxford Dictionary defines AI as:
The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.
While introducing human-like intelligence to machines, there are several goals of AI that developers want to achieve:
- Knowledge representation
- Natural language processing
- Ability to move and manipulate objects
To achieve these goals, developers rely on statistical methods, computational intelligence, and traditional symbolic AI.
What is the history of AI development?
As an academic discipline founded in 1955, AI was founded on the assumption that human intelligence “can be so precisely described that a machine can be made to simulate it.” There have been ethical (e.g. killer bots) and economic (e.g. mass unemployment) concerns about AI development.
AI development trajectory did not have a smooth ride with new capabilities developed with new approaches and disappointments fueled by unrealistic expectations.
While the field of AI expands and computers become more capable, businesses can automate increasingly more tasks. Now, machines can successfully understand human speech, out-compete humans in strategic games like chess and Go, complex computer games like StarCraft II, route deliveries and handle autonomous vehicles.
Have we reached peak AI?
Although AI has been a popular field since the last decade, we might not have reached its peak. According to Google Trends, we observe an increasing trend as we continue to hear about emerging AI technologies like XAI, transfer learning, and AutoML. We see a more explicit growth in the interest in AI after 2016. Here are our observations about the increasing interest in AI.
Advances in deep learning contribute to the increasing trend
Deep learning is one of the main drivers of the increasing trend of AI. This technology had a growing trend since 2012 and reached a peak in 2018. Deep learning algorithms, like neural networks, might help us discover more about AI and contribute to the increase in the AI trend in a few years.
We believe that the recent decrease in the trend is because people start to shift their interests to new technologies. However, this decrease does not mean that we are all done with deep learning.
Similar growth is observed in the emerging technologies after 2016
When we look at emerging technologies, we observe an evident growth after 2016. The similar growth can also be seen in the AI trend in the same period. This situation shows that the interest in emerging technologies does increase with the popularity of AI.
Before 2016, we see that XAI, transfer learning, and AutoML have stable interests at similar levels. As the interest in these technologies grows afterward, we observe that XAI has a smaller increase in popularity, compared to transfer learning and AutoML.
On the other hand, self-supervised learning hasn’t been identified as a topic by Google Trends yet. We observe that this technology started to gain interest by 2018, and there is an increasing trend since then. Although the popularity of self-supervised learning is much lower than other emerging AI technologies, this might be because it is a newer technology than others. We can expect that its popularity will increase in a few years.
Investments in AI by governments and investors confirm that numerous market participants believe in the long term potential of AI. Ishan Manaktala, CEO of Symphony AyasdiAI, shares the following about governments’ investments on AI.
China has invested over $140 billion, while the UK, France, and the rest of Europe have plowed more than $25 billion into AI programs. The US, starting late, spent roughly $2 billion on AI in 2019 and will spend more than $4 billion in 2020.
The total AI investment numbers are higher than the government investments. Below, we have gathered different resources about how much spending has been and will be made. The AI market is expected to grow significantly by 2025. When we compare with the 2019 data, most resources estimate the market to be 8-10 times larger in 2025.
- According to the National Venture Capital Association, $18.5 billion of investment is made on US-based AI companies in 2019. This number was $16.8 billion in 2018.
- Crunchbase indicates that AI startups raised $24.7 billion in 2019 while it was $21.5 billion in 2018.
- A 2019 report of KPMG expects that total spending in AI, machine learning, and RPA will reach $231.9 billion by 2025.
- Research by ResearchandMarkets.com shares that the total spending on AI in India will increase from $1.2 billion to $11.8 billion between 2019-2025.
- Tractica estimates that the worldwide AI software market will reach $126.0 billion by 2025.
What are the different types of AI?
AI can be categorized in different ways by:
- Application scope: Narrow AI and general AI (AGI). Narrow AI systems, the AI systems in use today, are focused on a specific task like extracting data from documents. Achieving human-like general intelligence (AGI) that can be applied to a diverse set of tasks is among the field’s long-term goals. Prominent AI scientists expect AGI to take a few more decades to be realized.
- Application areas: Narrow AI applications are specialized in different application areas such as Natural Language Processing (NLP), computer vision, robotics wtc.
- Technical approaches used in development: AI field draws upon computer science, information engineering, mathematics, psychology, linguistics, philosophy among other areas. As a result, there are numerous technical approaches such as genetic algorithms, artificial neural networks, decision trees, deep learning etc.
What are the recent developments?
By benefiting from algorithms like neural networks, AI technology is growing to provide more accurate insights for businesses. The recent developments in AI make agents achieve specific human capabilities, like learning, vision, and language processing.
Advances in computer vision
The main objective of computer vision is to help computers to understand and to implement human visual perceptions. In this field, the most exciting development is the autonomous cars. Computer vision is one of the key technologies that allow self-driving cars to operate without human drivers in the vehicle in specific pilot areas like Chandler, Arizona. However, these vehicles still rely on remote operators when they can not decide what to do with enough confidence, for example when they encounter a situation they have not faced before.
An essential aspect of self-driving cars is the casualties they will prevent, and the time they will save.
Advances in Natural Language Processing (NLP)
While NLP is used in many areas, including article summarization, instant translation, spam detection, and information extraction, recent developments help virtual assistants to behave more human-like. Now, AI-powered digital human assistants are starting to be integrated into companies to improve business efficiency thanks to advances in NLP technology. You can read more about this from our AI avatar article.
Explainable AI (XAI)
One of the most commercially successful AI approaches is deep learning. However, deep learning is like a black-box model and it is hard to explain why deep learning models make the predictions they do. Businesses naturally don’t prefer to take action without understanding their reasons.
To solve this issue, Explainable AI (XAI) tools are starting to emerge. XAI tools provide visualization and what-if analysis for humans to perceive how these tools reach specific conclusions and make AI more understandable for humans. Feel free to read our detailed guide about XAI if you are interested.
Deep learning is a broader family of machine learning techniques based on artificial neural networks. While the human brain inspires this algorithm, it is commonly implemented to other technologies like computer vision, natural language processing, social network filtering, and bioinformatics to boost their performance. A real-life example would be the face recognition technology in our phones. Face unlock function is available by deep learning algorithms applied to computer vision.
Reinforcement Learning (RL)
While humans learning from their experience is taken as an example of this development, RL is widely used in robotics today. Robots, chatbots, and game engines like AlphaGo are learning through their experience and are improving their performance during their learning process. As an example, you can see a robot taught itself how to walk by itself.
Transfer learning enables users to benefit from the knowledge gained from a previously used AI model for a different task. As an example, an AI model that is developed for English speech recognition can be used for German speech recognition. Using a pre-trained model will save time rather than starting from scratch. This method can be applied when:
- creating a new learning process might take too long
- there is no enough data for that specific task
Self-Supervised Learning (Self-Supervision)
Self-supervised learning, also known as self-supervision, is an autonomous learning technique that helps data scientists understand human intelligence further. The learning model trains itself by taking one part of the data to predict the other part. As a result, it generates labels autonomously and eliminates the necessity of humans to label data. Today, this learning model is used to improve computer vision and natural language processing algorithms in tasks like 3D image rotation, distortion, or speech analysis.
What are the trends shaping the future of AI?
The future of AI will be shaped by the major AI use cases that will unlock significant economic value. Some of these use cases are AI assistants, autonomous driving, and AI-powered medical diagnosis. Additionally, AI technology will be used in combination with other emerging technologies like Mixed Reality (XR), IoT, and cloud computing. We can categorize the factors that will drive the future growth of AI under three main categories.
Availability of more data
Data availability has been growing exponentially and is expected to continue. With the increasing amount of data, AI agents can learn more and understand human capabilities better.
New AI algorithms are driving the growth of AI. However, scientists need to rely on more complex AI algorithms to improve the performance of their solutions. The main problem is that the interpretability of these algorithms reduces while the accuracy increases.
Improved computing power
If the past is any indication, even if new techniques are discovered, it may be a long time until we have enough computational power to use them. After all, the human brain is estimated to have more computational power than the best supercomputers, and we don’t precisely know the computational capacity of the human brain yet.
AI technology requires greater computing power to run the latest solutions with high performance. With today’s computing technology, it might take too much time for specific AI algorithms to handle high amounts of data. While tech giants like Amazon are producing AI chips to accelerate this process, advances in quantum computing might bring a step-change. You can read more about AI chips and quantum computing from our related articles.
Besides these significant steps, even smaller advances in AI can unlock economic miracles. Self-driving cars are already on the roads, and if we can improve their effectiveness by a few percentage points, they could add hours to many commuters’ lives. Even though such breakthroughs are merely incremental technical advances, they will require significant investment, engineering, testing, regulatory approvals.
For more on the future of AI, feel free to read our dedicated research on the topic.
What are the AI application areas by business function?
AI is becoming more integrated into our lives, with more AI use cases emerging. Businesses are now introducing AI-powered solutions to their systems to generate more output/revenue, reduce their costs, and improve their customer satisfaction.
Companies can combine customer data and AI-powered tools to anticipate their customers’ next move and improve their journey. To do that, they can use AI to understand the market better, create unique content, and perform personalized marketing campaigns. With customer data, AI can provide accurate insights and suggest smart marketing solutions that would directly reflect on profits.
Unlike marketing, the sales function has always been numbers-driven. With the explosion of sales data and computational power, AI is set to increase further how data-driven the sales function is. AI-powered sales use cases can lead to conversion rates, reduced costs, and more accurate sales predictions.
AI algorithms can provide beneficial insights and help companies for better decision making. From production to sales, AI tools can identify the main bottlenecks of your business, provide KPI metrics, and see your strengths and weaknesses.
Customer satisfaction is a vital metric for companies, and they need to understand customer processes well to keep them satisfied. By learning from customer experiences, AI tools can offer companies a wide range of solutions and help them to provide better customer experience in the future.
Since most AI solutions rely on data, data is critical for the success of a company’s AI initiatives.
AI constitutes an essential part of the tech that companies use. AI algorithms power many automation and robotics applications. Companies can advance higher efficiency rates and improve their performances by integrating AI tools into their tech.
HR teams can benefit from AI during hiring processes and performance management. AI tools can interview and assess candidates, track employee performance, find ways to improve performance and serve as digital assistants.
There are AI solutions for every business function. For a detailed analysis of 100+ areas of AI application, feel free to read our article on the topic.
What are the AI application areas by industry?
As the range of AI use cases expands, businesses can uncover industry-specific applications of AI. These applications include self-driving cars, AI-powered credit scoring, etc. Here are some of the industries where AI technology can be adapted:
How to learn more?
Conferences are a more interactive way to learn more about AI. Feel free to check out our prioritized, sortable, filterable list of AI events for business users.
You can also read our other articles about AI, if you want to learn more. These articles can also interest you:
- Potential timing of Artificial General Intelligence/Singularity
- Advantages of AI according to top practitioners
- AI in Automation: Which tasks can we automate?
- AI chips: Guide to cost-efficient AI training & inference
- AIOps: Guide to Integrate AI into Your IT Operations
- AI Avatar: In-depth Guide for Businesses
If you have questions about how you can integrate AI into your business, don’t hesitate to contact us: