read more
Artificial intelligence fed by AI-generated data may 'break down'
The introduction of AI will certainly not lead to mass layoffs of employees, because controlling the technology is necessary to avoid mistakes. Moreover, artificial intelligence is unable to acquire certain human characteristics necessary for this work. Despite this, AI may take over 1/3 of the market in the next 4-5 years. Then the savings will start to be noticeable. However, customers will feel the price increases sooner, because the introduction of modern technologies involves additional costs. However, it will stabilize over time, says Łukasz Zieliński.
Fifty fifty
Artificial intelligence is machine learning on hundreds of thousands of examples. Therefore, it should be considered both in terms of opportunities and threats. More or less half and half. The former include improving market analysis and creating research tools (e.g. in-depth questionnaires). AI is already accelerating qualitative IDI research by reading responses and their context in real time.
Responsible use of AI will make the tools created more accurate, which will translate into shortening the time needed to conduct projects in the long run. In a short period of time (3-5 years), you can expect work on tools, their improvement and effective use. It will be like introducing new software. There will be high costs at the beginning that cannot be avoided, but they will even out over time.
However, bot farms pose a serious threat. They will constantly generate significant costs because they will have to be eliminated or avoided. Additional verification will be needed because not everything can be done accurately. Research is human-driven and AI is automation. It has already come to the point where booties pretend to be respondents in CAWI and CAPI surveys. There are currently several dozen variables that booty uses. The fight against them will involve continuous creation and improvement of security measures. Because they learn from mistakes and adapt algorithms, bypassing, for example, open questions.
Bot farms can perfectly falsify various parameters so that the audit or interview looks like a real one. This will have to be fought. And this, in turn, will incur costs. And this will be a fight not for a moment, but permanently, and unfortunately it will be costly. However, there is no other way here.
It is not obvious whether improving research processes will compensate for the workload and maintaining high algorithm efficiency. Opinions on artificial intelligence are divided. Some researchers see it as a great opportunity, while others believe it will ruin their business. AI will certainly enter the research market, but it will not replace people because it will not adopt some of their features. It's about experience, knowledge and even empathy.
Ultimately, AI for the industry means more or less as many advantages as disadvantages. But this conclusion should not impact investing in AI. In fact, it should be done in a responsible way so as not to be outdone by the competition. You need to participate in building new solutions so as not to be left at the end of the chain.
Evolution, not revolution
If the industry approaches the topic well, it will learn to properly use AI support. Then it will be able to automate many processes, under the watchful eye of a human. There's no way it's going to take people's jobs because someone will have to handle it. Otherwise, mistakes will be made. AI will certainly change the research market, but it will still be just a tool. An evolution rather than a revolution is coming. It can be predicted that within 4-5 years AI will take over 1/3 of the market.
AI can only change the CATI testing market to the extent that humans are willing to answer AI questions. Hotline chatbot evaluations can provide information on whether an AI telephone interview will be effective. At first, respondents may treat such a conversation as a kind of attraction, but later the situation will reverse. There will be weariness, and there will also be frustration, because AI is not a human. He approaches the matter rigidly and, above all, task-oriented.
In addition, AI will help in the preparation of research tools – from an interview questionnaire to a script. He already replaces qualitative researchers in certain tasks. Develops IDI tools that adapt to responses and create a narrative report almost simultaneously. Currently, the moderator is leading one conversation. However, a tool using AI engines can conduct many interviews at the same time. The time savings are enormous, provided that a person has control over it.
At the beginning, the most advanced tools will be used by the largest companies on the market, due to their financial resources. Over time, this technology will be used by the entire market to some extent, because it will be a kind of compulsion. Of course, in the first phase, the implementation of AI can be used for marketing purposes towards customers, but it is difficult to clearly predict how they will evaluate it. They are more likely to look at it from a distance or treat it as some kind of curiosity. It can be predicted that retail will soon want to use artificial intelligence.
The beginnings can be difficult
At the beginning, clients of research companies may be quite skeptical about the topic, mainly out of concern for quality. And the point here is not that AI will make any significant mistakes, but that some smaller mistakes will not be noticeable, which will somehow affect the assessment of the research results.
For a long time, customers may indicate that these types of machines and tools are still learning, and they need high-quality results and do not have time to wait. And in a sense they will be right. They will certainly also express a degree of distrust in the conduct of individual research processes. They may also indicate that such a tool will be easy to deceive.
Already today, customers claim that respondents will not want to talk to AI because it looks quite artificial. They also add that when people come into contact with artificial intelligence, they also learn, mainly how to lie to the machine to get the purpose of the conversation faster.
It may also be that customers will agree to research combined with AI, but will expect lower prices. It was similar in the case of CAWI research. There was a moment when service users were skeptical about online research, claiming that it was much weaker than CAPI. However, this time passed quickly because customers realized that the quality was good and the final price was better. The same may apply to the use of AI tools.
However, the introduction of AI will not force research companies to lower research prices, because investments in new technologies are expensive. We will need to constantly invest and maintain technology, as well as create anti-bot protection. The resulting costs would have to be passed on to customers, and they would start forcing discounts or try to do so, but without much success. The prices of research services will increase because the industry will want a return on investment. Savings can only be expected in 4-5 years.
The author of the commentary is Łukasz Zieliński, an analytical and research market expert from the international company SYNO Poland.
