Artificial Intelligence is Becoming Better than Human Expertise

It is believed that at the turn of the 20th and 21st centuries, there were three distinct waves of AI technological development. The first wave was an attempt to create general intelligence with the help of the formal rules of logical thinking. The second was the development of expert systems. Now we are on the crest of the third wave, the main feature of which, computer learning, is already successful in helping artificial intelligence bypass top professional experts in the effectiveness and quality of decision making on a massive scale.

Researchers have learned how to create systems that improve the algorithms themselves and then receive new data. This invention has created a new industry. The introduction of AI by 2030 will boost worldwide GDP by 14%, or approximately $15.7 trillion, according to PwC. They note that this is more than the current total industrial output of China and India combined. According to the results of a study by Teradata, the vast majority of large companies (80%) are investing in artificial intelligence technologies, and according to forecasts from Gartner, by 2020 they will be present in almost all new software products and services.

What is the Advantage of Artificial Intelligence?

Until recently, the only way to solve complex problems was by working with specialized experts. On average, their experience and abilities are superior to those of ordinary people, but there are many factors that can influence their effectiveness.

There are natural limitations to the productivity of the human brain. We can turn to experts for help in assessing the potential of investing in a startup, but what if we need to analyze the potential of not just a single company, but several thousand? There is no one expert, or even a group of them, that would be able to efficiently process such a high volume of information, while the use of AI makes issues like these fully resolvable. For example, Squilla Capital now uses Artificial Intelligence and Big Data to analyze more than 8,000 startups, using such metrics as web and social media analytics, trading data, blockchain data and others.

Danil Myakin, co-founder of Squilla Capital, states:

People always remain biased and emotional, regardless of whether they are aware of it or not. Everyone knows that people make mistakes. In my opinion, it is much worse to rely on guesswork and intuition, and not on data and statistics. We are now ready to give investors the means to assess tremendous amounts of data without data analytical skills and heavy tools. Do not guess. Use data.

Experts, like all people, are subject to cognitive distortions, which fundamentally affect the quality of the result. Prejudices, false assumptions, the desire to simplify, ignoring contradictions and, of course, the emotional component — all this affects the result, and all these weak points can be avoided with an introduction of AI in analytical processes. Moreover, experts’ opinions are too difficult to interpret unequivocally; they protect themselves and avoid specific details and clear phrases, which also exclude algorithms that produce dry numbers.

Warren Buffett, one of the world’s most successful investors, said:

You don’t need to be a rocket scientist. Investing is not a game where the guy with the 160 IQ beats the guy with a 130 IQ. Rationality is essential. Even if you do have an IQ of 160, you should just give away 30 points to somebody else. You don’t need a lot of brains to be in this business. What you do need is emotional stability. You have to be able to think independently.

In the coming years, technology giants such as Google, Facebook, Uber, and Apple will earn their profits primarily from the introduction of AI. With the help of computer learning, they will be able to solve the complex problems of creating computer vision, motion control modules, unmanned vehicles, speech recognition, and organizing and providing access to information.

For the smaller participants in the third wave, there are also excellent prospects. The potential lies in the application of artificial intelligence to specific local tasks. I will offer a few examples where the potential for using technology is the most obvious.

Artificial Intelligence vs Professional Experts

1. Finance

The financial industry remains the leader in the number of relevant projects. Here, technologies allow the industry to reduce costs, minimize risks, prevent fraud, verify borrowers and evaluate their solvency, as well as make predictions, among other tasks.

Earning more than the average in the stock market is almost impossible; even the most talented investors on Wall Street are not known for their consistency. Traders and hedge fund managers cannot be competitive compared with AI, their problem is that they are only human, while all the decisions made by robots are based strictly on data and statistics.

A research team from the University of Erlangen-Nuremberg in Germany has developed a number of algorithms that use historical data from markets to replicate real-time investments.

One of the models allowed for a 73% return on investment annually from 1992 to 2015, taking into account transaction costs. This compares with a real market return of 9% per year. Profits were particularly high during the market shocks of 2000 (a 545% yield) and 2008 (a 681% yield), which proved the increased efficiency of quantitative algorithms during periods of high volatility, when emotions dominate the markets.

Eurekahedge’s January study of 23 hedge funds using artificial intelligence showed that they had much better results than those managed by people.

Over the past six years, these funds have achieved an annual yield of 8.44% compared with conventional funds, which ranged from 1.62% to 2.62%. The authors of the study associate the dominance of artificial intelligence in the industry with the fact that it constantly conducts repeat testing, and does not just accumulate data. This may also be due to the shortcomings of traditional quantum approaches and the use of trading models built using unprofitable backtests based on historical data that are not capable of generating profit in real time.

The Medallion Fund from Renaissance Technologies, which uses quantitative stock market analysis techniques, boasts some of the best performance in the history of investment. For 20 years, the fund was able to generate returns of 35% annually. This means that if you invested $10,000 in 1997, today you would have $4.04 million in your hands.

Some companies use artificial intelligence to ensure profitability via algorithmic trading. In just a few minutes, The Sentinent Technologies Fund can simulate 1800 trading days, pitting trillions of virtual traders against each other.

2. Software Creation

Google’s artificial intelligence learned how to create machine learning software that is more efficient than human-created systems. This was written about in Wired. The system needed to recognize several objects in one image. The computer-trained algorithm recognized up to 43% of them, while the best of those created by people — only 39%.

As Google’s CEO, Sundar Pichai, put it at the company’s latest presentation, the company has set itself the objective of helping artificial intelligence developers with the new system. AutoML, he said, is necessary not to replace human developers, but to aid them during the initial setup of self-learning systems.

Sundar Pichai stated:

Today these are handcrafted by machine learning scientists and literally only a few thousand scientists around the world can do this. We want to enable hundreds of thousands of developers to be able to do it.

In fact, developers are now at the point where artificial intelligence itself has surpassed man in the creation of artificial intelligence.

3. The Legal Practice

The startup LawGeex conducted a study comparing the effectiveness of lawyers with artificial intelligence in assessing the quality of non-disclosure agreements. The study involved 20 lawyers with dozens of years of experience in companies such as Goldman Sachs, Cisco and others, as well as LawGeex’s proprietary AI platform. Participants evaluated the risks contained in five different NDA agreements by searching for 30 specific legal points.

According to the results of the study, AI showed an average accuracy of 94%, whereas people achieved 85%. At the same time, the AI’s maximum accuracy was 100%. The lawyers — 94%. On average, people required 92 minutes to accomplish the task, while the AI analyzed all the documents in 26 seconds.

4. Medicine

A research team from the National Institutes of Health and Global Good has developed a method based on AI which will help analyze digital images of a woman’s cervix and accurately identify pre-cancerous changes that require medical care

Mark Schiffman, senior author of the study, stated:

Our findings show that a deep learning algorithm can use images collected during routine cervical cancer screening to identify precancerous changes that, if left untreated, may develop into cancer. In fact, the computer analysis of the images was better at identifying precancer than a human expert reviewer of Pap tests under the microscope (cytology).

In general, the algorithm worked better than all standard screening tests in predicting all cases diagnosed in the Costa Rica-based study. Automatic visual assessment revealed precancerous disease with greater accuracy (AUC = 0.91) than human examination (AUC = 0.69) or conventional cytology (AUC = 0.71). An AUC value of 0.5 indicates a test that is no better than a random one, while an AUC value of 1.0 represents a test that has perfect accuracy in detecting disease.

5. Military Use

The new ALPHA artificial intelligence was able to defeat a professional expert in simulating air combat. The AI itself was developed by graduates of the University of Cincinnati for research purposes.

ALPHA was tested by retired US Air Force Colonel Gene Lee, who has extensive combat experience and is an expert in air combat tactics. The AI, according to the developers, has provided a breakthrough in the field of control systems, whose work is based on the application of fuzzy logic. Colonel Lee was not only unable to defeat the program in a virtual air battle, but also was “killed” every time he fought long battles with it. According to the military man, he was surprised that ALPHA seemed to anticipate his intentions and instantly react to his actions.