The exciting future of artificial intelligence

2021/ 13/09

Where is artificial intelligence going? What developments can be expected in the short and long term? Is it possible to develop artificial intelligence systems that match the performance of the human brain, or is this just fiction? What technological developments will allow the development of self-driving cars or robotic surgery? In our article below, you can find out! 

 

 

What technologies can we count on for the development of artificial intelligence?  

Artificial intelligence systems have so far relied mainly on increased computing power, more data availability, better algorithms and better tools to sustain progress. There is potential for dramatic improvement in all 4 areas, but they are difficult to position well in time. 

 

The evolution of computing power 

Deep learning relies on computing power to solve more complex problems. With current technology, learning can take too long to be profitable, so there is a need to improve computing power. With new computing technologies, companies will have artificial intelligence models that can learn to solve more complex problems. 

 

AI-based chips   

Even the most advanced CPU alone may not improve the efficiency of the AI model. To apply AI in applications such as computer vision, natural language processing or speech recognition, companies need powerful CPUs. AI-enabled chips are a solution to this challenge, as these chips make CPUs "intelligent" to optimise their tasks. As a result, CPUs can work individually on their tasks and improve their efficiency. New AI technologies require these chips to solve complex tasks and perform them faster, so companies such as Facebook, Amazon and Google are increasing their investments in AI-enabled chips. 

 

Progress in algorithm design  

While AI capabilities are evolving rapidly, the algorithms behind AI models will also evolve. Advances in algorithm design will allow AI to be more efficient and accessible to more people with less technical knowledge. The followings are the most prominent advances in AI algorithm design. 

 

Transfer learning  

Transfer learning is a machine learning method that allows users to use a previously developed AI model for a different task. For example, an AI model well trained to recognise different cars can be used for trucks. Instead of starting from scratch, the knowledge gained about cars will be useful for trucks as well. 

 

Reinforcement Learning (RL)  

Reinforcement learning is a subset of machine learning, unlike traditional learning, RL does not look for patterns to make predictions. It makes sequential decisions to maximize the method and learns from experience. Today, the most common example of RL is Google's DeepMind AlphaGo, which beat the world's number one Go player, Ke Jie Ke Jie, in two consecutive matches. In the future, RL will also be available in fully automated factories and self-driving cars. 

 

Self-supervised learning (self-supervision)  

Self-supervised learning (or self-supervision) is a form of autonomous supervised learning. Unlike supervised learning, this technique does not require a human to label the data, since the labeling task is performed by itself. According to Yann LeCun, Facebook's vice president and chief artificial intelligence researcher, self-supervised learning will play a critical role in understanding human-level intelligence. Although it is now mostly used in computer vision and NLP tasks such as colouring images or language translation, it is expected to be more widely used in our everyday lives. Some future applications of self-supervised learning will be in healthcare (endoscopy and robotic surgery) and autonomous driving (estimating terrain roughness). 

 

What can we expect from the development of artificial intelligence in the coming years?  

 

There is virtually no major industry that has not already been impacted by modern artificial intelligence (or more precisely, "narrow AI", which uses data-driven models to perform objective functions, often falling into the category of "deep learning" or machine learning). This has been particularly true over the past few years, as data collection and analysis has accelerated significantly thanks to robust IoT connectivity, the spread of connected devices and increasingly fast computing.  

According to the AI Index, the number of active AI startups in the US grew by 113% between 2015 and 2018. Thanks to recent advances in "deep learning", AI now can power search engines, online translators, virtual assistants and many marketing and sales decisions. PwC reports that AI will contribute $15.7 trillion to the global economy by 2030. So let's take a look at how the future of AI could make this possible: 

 

1. The future of AI in healthcare  

The healthcare sector is also set for dramatic change. AI has the ability to recognise illness based on symptoms; even if you don't go to the doctor, as it can read data from your fitness watch/medical history to analyse trends and recommend appropriate medication or even call an ambulance in case of emergency (as the Apple Watch can do today).  

AI-enabled robots will be able to perform complex surgeries with a high degree of accuracy. In addition, AI will increase the efficiency of healthcare professionals to better understand the health background of the people they care for, enabling a more holistic approach and therefore the ability to provide better feedback, guidance and support to maintain health.  

Google's 'deep mind' has already been able to partially beat doctors in detecting deadly diseases such as breast cancer. The time is soon when artificial intelligence will be able to recognise common diseases as well and make appropriate drug recommendations. On the other hand, one of the long-term consequences of this could be that there will be less need for doctors, which could lead to fewer job opportunities. 

 

2. AI in education  

The development of a country depends on the quality of education of its young people, thats why it is important to recognize the fact that AI is expected to transform the classical way of education as well. The world no longer needs skilled workers in manufacturing industries, which have been mostly replaced by robots and automation. The education system can become very efficient, tailored to the personality and skills of the individual. It would give smarter students a chance to show their potential, and lower performers a more effective way to cope with the learning material. 

In addition, textbooks will be digitalised using artificial intelligence, virtual tutors will assist human tutors, and facial analytics will assess students' emotions to help determine who is struggling or bored, and therefore better tailor the learning experience to individual needs.   

The right education can enhance the strength of individuals/nations; on the other hand, its misuse can lead to devastating results. 

 

3. Artificial intelligence in finances 

Quantifying growth for any country is directly linked to its economic and financial situation. Because AI has huge potential in almost every field, it has great potential to boost the economic health of individuals and a nation. Today, AI algorithms are used in equity fund management.  

An AI system would be able to take many parameters into account, while also finding the best way to manage a financial fund, performing better than a human manager. AI-driven strategies in financial markets will therefore change the classical way of trading and investing. This could be destructive for some fund management firms that cannot afford such tools and could have a major impact on business, as decisions would be made quickly and abruptly, making competition fierce and constantly tight. 

 

4. Smart cities  

Cities are complex systems, but stripped down to the basics, they are made up of people and communities interacting with objects such as roads, buildings and spaces in a wide variety of environments and contexts. When cities are called smart cities, it essentially means that different sensors collect data in order to extract information and use it to effectively manage these objects. One example of this is minimising asset retirement and increasing the life expectancy of assets. These lead to higher performance, lower management costs, happier asset users and better sustainability. 

 

5. Artificial Intelligence in military and cyber security  

AI-enabled military technologies have built autonomous weapon systems that do not require humans at all to operate, making them the most reliable way to enhance national security. In the near future, we may see military robots that are as intelligent as a soldier/commando and will be able to perform certain tasks.  

AI-enabled strategies would increase the effectiveness of missions and provide the safest way to carry them out. The concern with AI-enabled systems is that the way the algorithm is executed is not fully explained, as deep neural networks learn faster and continuously. It could lead to devastating results if the program makes bad decisions or falls into hostile hands. 

 

6. Self-driving vehicles  

Nowadays, everyone is talking about "self-driving vehicles", and Tesla has already achieved autonomy level 2.0. Autonomous driving is one of the most important applications of artificial intelligence.  

Autonomous vehicles (AVs) are equipped with multiple sensors, such as cameras, radars and lidar devices, to help them better understand their surroundings and plan their routes. These sensors generate huge amounts of knowledge.  

As autonomous vehicles become more common on the roads, taxi services like Uber and Ola will become driverless, which would change the way the transport industry works. The market for AI-driven autonomous vehicles is forecast to reach $127 billion by 2025.  

 

7. Entertainment  

OTTs like Netflix and Amazon Prime are already growing their user base rapidly. Intelligent algorithms will be ready to come up with the most effective marketing and advertising solutions. With AI, predictive analytics will be used to make all marketing processes faster. In the future, AI will be able to predict not only our preferences but also our moods and display content accordingly. 

 

8. Production  

In the near future, production will be fully automated. Artificial intelligence will also optimise manufacturing supply chains, helping companies to anticipate market changes. This information is invaluable for manufacturers, allowing them to optimise recruitment, internal control, energy consumption and therefore raw material supply.  

The manufacturing processes enabled by artificial intelligent systems would not only be able to perform the necessary processes, but would also be able to monitor, improve and control the quality of products without human intervention. According to Marketsandmarkets report, the Artificial Intelligence in Manufacturing market is expected to grow from USD 1.0 billion in 2018 to USD 17.2 billion by 2025, at a CAGR of 49.5% during the forecast period. 

 

The distant goals  

The ultimate goal of artificial intelligence - to have a machine with the general intelligence of a human - is one of the most ambitious goals ever put forward by science. It is comparable in difficulty to other major scientific goals, such as explaining the origin of life or the universe, or discovering the structure of matter.  

The human brain is in fact very far removed from current models of artificial intelligence, suggesting that the so-called singularity - artificial superintelligences based on copies of the brain that far surpass human intelligence - is a prediction with very little scientific basis.  

All research efforts in the field of artificial intelligence have been directed towards the creation of specialised artificial intelligences, and the results have been spectacular, especially in the last decade. This success is essentially due to a combination of two elements: the availability of huge amounts of data and access to the sophisticated computing tools needed to analyse them.  

 The most complex ones are those capabilities that require interaction with an unrestricted and unprepared environment. The design of systems with such capabilities requires the development of several areas of artificial intelligence. Furthermore, developmental robotics may be the key to endowing machines with common sense, in particular the ability to learn the relationships between their actions and their impact on their environment.  

Ultimately, intelligent future artificial intelligences will never be the same as human intelligence: the mental development required for complex intelligence depends on interactions with the environment, and these interactions depend on the body - in particular its sensory and kinetic systems. 

Sources: Educba, AI multiple, Openmind BBva, Built in, Techvidvan, Technology Review

Try our AI-based speech recognition application for free!

Speed ​​up your work with artificial intelligence! With the help of Alrite, you can easily create Hungarian transcriptions and video captions for dictated or previously recorded audio and video materials. The application offers the ability to store files, edit and share transcriptions and captions, and perform advanced search options.

More information