Tali Soroker is a Financial Analyst at I Know First
- אלגוריתמים משמשים באופן יומיומי לניתוח התנהגות אנושית וקבלת החלטות.
- חברות משתמשות באלוגריתמים הללו על מנת להגיע ללקוחות רלוונטים. ולהגדיל את קהלי היעד שלהם.
- בינה מלאכותית הוכחה ביעילותה למעבידים המחפשים עובדים מוצלחים ובעלי יכולות מגוונות.
- שימוש באלגוריתם לניתוח תהליך מחשבת האדם, יכול לעזור למשקיעים לבצע החלטות השקעה מוצלחות.
Warren Buffett, the most successful investor in the world, once said, “beware of geeks with formulas.” Looking towards to future of the job market, in response to the rise of technology and the development of algorithmic systems and artificial intelligence, some consequences of the advancements in algorithmic formulas are troubling. Researchers are estimating that within the next decade or two, close to 50% of American jobs will be automated. This is including not just production and manual labor, but research and analyst positions as well. Artificial intelligence is capable of more today than many people can imagine. AI is, in many ways, being used to characterize and analyze human behavior. It’s fair to be wary of formulas in a computer system that is analyzing human behavior only to place people in boxes. Algorithms are now being used to optimize search results based on what some are calling discriminatory parameters, and they are even being used to assist in the hiring process at some companies. Algorithms can also be used, not to do analysis on separate individuals, but to analyze human behavior as a whole.
Search Engine Optimization (SEO) Algorithms Become Discriminatory
It shouldn’t be new to anybody that the websites that we use are working to show us the content that we are most interested in. There are algorithms running behind the scenes of sites like Google, Facebook, and even your favorite dating website. At first glance, this seems wholly beneficial. When Facebook first introduced the News Feed in late 2006, there was already a crude algorithmic system in place to determine the ordering of posts. What started as a simple and almost arbitrary ranking system designed by Facebook’s engineers, has become an incredibly powerful tool that ranks posts based on a wide range of parameters. If for example, somebody is on the Facebook app on their phone and they have a slow data connection, Facebook’s News Feed will show them fewer videos as their buffering time would be slowed. If somebody comments on their friend’s status, “congratulations,” Facebook will boost that status because it assumes the subject of the post is a noteworthy life event.
It’s not surprising that Google has a similar algorithm behind it’s iconic homepage, optimizing the search process. Google’s algorithms rely on hundreds of signals to provide users with the most relevant results at the top of the list. Some of these signals include terms on websites, the freshness of content, and your region in the world. The signals and criteria that the algorithms use in order to determine which results best answer the search inquiry are all relatively straightforward. By learning more about how the algorithm processes information and what signals it’s looking at, companies are able to design their web content in order to optimize their ranking in Google search results. People say that any link past the first page of search results essentially doesn’t exist, and they’re right, so mastering SEO can be vital to the success of a company.
The idea that our search results are listed in order of relevancy and quality is amazing, and there is no doubt that this kind of technology is necessary in order for us to effectively use the internet as a source of information. Now, though, researchers are studying the performance of the algorithms and asking about the potential drawbacks of such techniques. For example, is it possible that the algorithms mirror and perpetuate the discrimination of minority groups? The answer, as they have found, is yes. Researchers from various organizations looked not only at the order of search results but at the advertisements that are shown in connection with the search terms and results. Researchers who looked at search results for CEO images found just 11% of the images in the search results were women. One researcher even reported that in a recent search the first woman to appear, on the second page of results, was the CEO Barbie Doll.
(Source: Google Images)
While women CEOs are, in reality, a minority of just 27% in American companies, the lower percentage of images shown in search results can perpetuate sexist ideas about women holding high-level management positions. The algorithm’s decisions about which advertisements to show and to which people is just as troubling at times. Advertisements for high-paying jobs have been found to be shown more frequently to men than to women, ads for arrest records and bail bond companies are more likely to be shown when one is searching for historically black colleges, and advertisements are often targeted towards neighborhoods based on the median income of its residents. Clearly, despite the algorithm having a seemingly unbiased approach to analyzing human behavior and interests, it is learning from our flawed system and propagating the same unequal treatment that minorities face in the community.
Algorithms are Determining a Person’s Character in the Hiring Process
The hiring process is stressful and, often, unpredictable for both the interviewer and the interviewee. Walking into a job interview, it’s hard to know what exactly the employer is looking for, someone who is qualified and capable of completing the necessary day-to-day tasks or someone who makes them laugh and reminds them of their old college roommate. Companies recognize the discrepancies that exist in the hiring process as well as the lack of diversity that exists in the workforce. Hoping to make a change in the way people get hired, some start-up companies are starting to automate the process using algorithms and artificial intelligence.
While there are those who voice concerns about removing the human element from hiring, the founders of these start-ups see this human element in the form of bias and partialities that have very little to do with one’s ability to do the job that would be required of them. By using company data and public data from sites like LinkedIn, these start-ups are matching companies with willing and able employees that previously may have been overlooked. Employers and people doing the interviews for job applicants want to feel a chemistry with potential employees, and in many cases they hire based on this chemistry or based on a “gut feeling,” but there’s little correlation between this feeling and the future success of the applicant. In fact, it often leads to the hiring of people who share characteristics, backgrounds, and friends with the employer. The new use of algorithms in the hiring process may be able to improve diversity in the workforce by ignoring the very human element that employers are afraid to lose.
Analyzing Human Behavior and Decision-Making as a Whole
Applying artificial intelligence to the financial industry takes a different approach entirely. Rather than focusing on individual characteristics or information, algorithmic processes are used to analyze human behavior and decision-making as a whole. Algorithmic analysis of financial markets is not about placing individuals in boxes, it’s about placing all people into the same box and then interpreting their reactions and interactions.
There is no single way to make trades in the stock market. Each investor approaches the market in a slightly different way. Some traders buy stock based on news stories, some buy stocks from companies that they admire or that have products that they like. Other traders look at fundamentals, at balance sheets, or do technical analysis. Regardless of the preferred method of trading, each individual investor has his or her own emotions affected the trades that they make. Whether the trader is making decisions based on fear or excitement doesn’t matter, what matters is the effect that these emotions have on the market. Mass psychology perpetuates the effect on the market, as people tend to follow the path of others and together they push the market prices in one direction or the other.
(Source: Google Images) – Risk-Return Indifference Curves
Though many traders use fundamental and technical analysis to determine which trades will be best for their stock portfolios, there is simply too much market data for any person to fully analyze and understand. Algorithmic systems, however, are able to take in years of market data and identify patterns that humans simply can’t see. Two important factors in market waves are people’s assumptions and emotional reactions to news stories. The stock market is not simply a numbers game, but a combination of company fundamentals and market noise. In order to make reliable forecasts of what future stock trends are likely to be, both sides of the market must be accounted for. Algorithmic trading combines analysis of stock trends and fundamentals with market noise and human emotion to output highly accurate market predictions.
Stock Market Predictions Based on Algorithm
I Know First has now developed a unique genetic algorithm that takes in 15 years’ worth of market data and then outputs market forecasts for 6 different time horizons spanning from three days to one year. This predictive algorithm combines analysis of market trends with market noise to improve the accuracy of its predictions. The stock market moves in waves driven by news stories and the interaction between different kinds of investors. I Know First’s system of algorithms finds trading opportunities between the different wavelengths.
Artificial intelligence has advanced to amazing lengths in the 21st century. Today, companies are using AI to analyze human behavior for a variety of reasons. Some wish to target advertising to consumers who are most susceptible to it, others want to use it to hire the most capable employees to help their company succeed. Many companies, in the financial sector, are using this same technology to predict future stock trends. Not all algorithmic processes are as transparent as others, just as each has its own unique design and purpose.