AI Search Engine
I Predicted Today’s AI Search Engine in 2017. I Was Right.
I Predicted the Future of the AI Search Engine. Nine Years Later, My Prediction Has Come True.
Long before ChatGPT. Long before the AI boom. I was already writing, publishing and discussing the future of artificial intelligence, big data and search engines.
In August 2017, I was featured in a Chinese technology publication in an interview titled:
Artificial Intelligence, Big Data and the Future of Search Engines.
I discussed AI-powered search, machine learning, neural networks, personalization, natural language and how search engines would evolve beyond traditional keyword-based technology.
I also published my book back then, the very first book about AI in digital marketing, with my prediction on the future of SEO and search: The Future of Artificial Intelligence in Digital Marketing: The Next Big Technological Break.
Looking back almost a decade later, it’s fascinating to see how many of these subjects have become central to today’s AI revolution.
Technology evolves. Ideas take time to become reality.
I’ve been exploring the intersection of AI, search technology and digital innovation since long before it became mainstream.
In August 2017, I shared my vision of how artificial intelligence would transform search engines in an interview with the Chinese technology publication Lieyunwang, later republished by Sohu. My prediction regarding AI and AI search engines was that traditional keyword-based searching would gradually give way to something much more sophisticated: intelligent search systems capable of understanding natural human language, interpreting intentions and delivering increasingly personalized information.
Nine years later, that transformation is no longer a distant technological possibility. It has become a defining direction in the evolution of search.
Looking back at my 2017 interview, what stands out is how many of the developments I discussed have since become important components of modern information retrieval.
What I Predicted in 2017
At the time, search engines largely relied on users entering keywords and receiving lists of links. Although search technology was already advancing, I believed that this approach would eventually become insufficient.
I argued that artificial intelligence could fundamentally change the relationship between people and search engines. Instead of requiring users to formulate queries around specific keywords, future systems would understand what people actually wanted to find.
My vision centered on four important developments.
Search engines would move beyond keywords.
I predicted that the traditional importance of keywords would gradually decline as search engines became better at understanding users’ intentions, interests and circumstances.
Rather than simply matching words, intelligent systems would need to interpret meaning and context. This would also change digital marketing, requiring businesses to understand their audiences rather than concentrate exclusively on keyword optimization.
Today, semantic search, natural-language processing and intent-based information retrieval are central to the evolution of search technology.
People would communicate with search engines naturally.
One of my most important predictions was that people would eventually no longer need to type traditional search queries.
I discussed personalized AI-powered search systems through which users could simply express what they wanted in ordinary language. Search engines would interpret those requests and provide relevant information.
This was a significant shift in thinking about the future of information retrieval. Search would become less dependent on the way people constructed queries and more dependent on the system’s ability to understand human communication.
Natural-language interaction and voice-enabled search have since brought this vision considerably closer to reality.
Search would become increasingly personalized.
I also anticipated that search engines would use contextual information to improve the relevance of their results.
A person’s interests, previous searches and individual circumstances could help determine which information was most useful. Rather than delivering identical results to everyone, intelligent search systems would increasingly adapt to individual needs.
Personalization has since become an important element of digital information services, with contextual signals such as location, language and user activity influencing the search experience.
Although privacy and data protection remain important challenges, the direction I identified in 2017 is clearly visible.
Different industries would need specialized search systems.
Another point I raised concerned the limitations of general-purpose search engines.
Businesses operating in specialized industries have their own terminology, information structures and professional requirements. A cybersecurity company, for example, needs to retrieve and interpret information differently from a general consumer searching for everyday information.
I argued that specialized search solutions would be necessary to address these differences.
Today, enterprise search, industry-specific information retrieval and domain-adapted AI systems are developing across sectors including healthcare, cybersecurity, finance, law and software engineering.
Nine Years of Technological Transformation
The developments since 2017 provide a clear historical context for my observations.
Google introduced BERT in 2019, improving its ability to understand the relationships between words and the context of search queries. This represented an important advancement in natural-language understanding.
The subsequent development of conversational search interfaces and generative AI-powered search features further changed how people interact with information.
By 2024, Google’s introduction of AI Overviews demonstrated how search was moving beyond presenting lists of links toward generating summaries and direct answers.
These developments reflect the broader technological direction I discussed in 2017: search systems becoming increasingly capable of interpreting meaning, understanding context and responding to human information needs.
What Has Changed, and What Remains Unfinished?
Not every aspect of my original vision has been fully realized. Personalized search still raises significant questions about privacy, accuracy and the responsible use of personal information.
Specialized search systems also continue to face challenges when dealing with complex professional terminology, fragmented datasets and information that requires deep contextual understanding.
Furthermore, intelligent search must do more than generate relevant answers. It must provide accurate, trustworthy and verifiable information.
These challenges do not diminish the significance of the original prediction. They demonstrate that the evolution of search is an ongoing technological process.
Looking Back at My 2017 Prediction
My August 2017 interview was not simply a discussion about incremental improvements to existing search engines. It explored a fundamental change in how people would interact with information.
I recognized that the future of search would depend increasingly on artificial intelligence, natural-language understanding, contextual interpretation and personalization.
Nine years later, these technologies are shaping the modern search experience.
I did not need to predict the name of a particular technology or product to identify where the industry was heading. What mattered was recognizing the underlying technological transformation.
The prediction I made in 2017 has become reality in ways that are now visible across the search industry.
The future of search was never simply about entering better keywords. It was about building intelligent systems capable of understanding what people mean.
And that is precisely the direction I identified nine years ago.
Original 2017 interview and historical sources
- Original interview republished by Sohu, 3 August 2017 (Chinese)
- Original publication: Lieyunwang
- The Future of Artificial Intelligence in Digital Marketing: The Next Big Technological Break (2017), by Maria Johnsen.
Maria Johnsen Interview: The History and Future of Artificial Intelligence, AI Search Engines and Big Data
Lieyunwang (WeChat: ilieyun), August 3, 2017
Editor’s Note: Maria Johnsen is a highly regarded AI search engine expert specializing in search engine technology research and development, as well as digital marketing.
How Did Artificial Intelligence Come Into Existence?
Over the past 200 years, philosophers have attempted to understand and address two fundamental questions: how human consciousness works and whether consciousness exists in beings other than humans.
Some philosophers have favored computational approaches that are now known as artificial intelligence (AI). John McCarthy, widely regarded as the father of artificial intelligence, defined AI as the science and engineering of creating intelligent machines. It is a branch of computer science that aims to enable computers, robots and intelligent software to operate in ways that demonstrate human-like intelligence.
However, artificial intelligence remains a subject of considerable debate. Some scientists believe that highly complex human behaviors, such as love, creative discovery and moral decision-making, are far beyond the capabilities of machines. Others even believe that artificial intelligence could become a means through which machines eventually dominate humanity.
Some regard AI as little more than a meaningless technological term that happens to form an important foundation of modern technology, helping solve numerous challenges in computer science.
Nevertheless, AI aims to solve problems by studying how the human brain thinks, learns and makes decisions.
What Role Does Big Data Play in Artificial Intelligence?
Big data refers to the enormous volume and variety of high-frequency digital information. Processing big data requires a combination of programming skills, domain expertise and statistical knowledge.
Big data comes from the internet, metadata (tags), translation systems and crowdsourcing platforms such as Amazon Mechanical Turk.
Big data is not particularly useful in itself. It is simply a collection of information unless you can find a way to use it effectively.
You cannot understand big data simply by looking at it. Nobody would sit down and read an entire telephone directory because doing so would serve little purpose. Big data only becomes meaningful when it is put to practical use.
For example, in the past, search engine programmers would not have examined data to generate results for queries such as houses, dogs or cats because users did not previously interact with these kinds of labels in the same way. Today, even my grandmother knows how to use tags and hashtags.
We can see the impact of big data every day. Facebook can process more than 40 billion user photographs. CERN’s Large Hadron Collider (LHC) generates enormous volumes of data, while Walmart processes more than one billion customer transactions in an hour, producing approximately 15 petabytes of data annually.
Can We Process Big Data Without Data?
No, you cannot.
Search engines do much more than simply index information. Today, people are increasingly concerned about search engines collecting data.
From a security perspective, this issue is often associated with copying sensitive files within systems. If data is written directly into an index, it may be more secure because indexes are stored as binary files that are difficult to browse and read directly.
Nevertheless, some of this information could potentially be exploited by companies or politicians.
What Advice Would You Give Regarding Privacy and Information Collection in Search Systems?
Innovation plays an extremely important role in the development of AI-powered search engines.
An intelligent search engine should be capable of understanding and handling unexpected situations originating from unpredictable sources.
Unfortunately, 99.9% of search engines today do not have such systems.
By unexpected situations, I am not referring to the programming term “throwing an exception” used in Java.
How Will Popular Search Engines Evolve Through Artificial Intelligence?
At present, Google’s AI system, RankBrain, has a very significant influence on the search industry, and this influence is expected to grow.
Other search engines, including Bing, Baidu and Yandex, have also begun incorporating AI processes into their search ranking systems.
RankBrain uses users’ historical interactions with individual search queries. This means that more information can be used to determine the most appropriate search rankings.
I believe these search engines will soon catch up with Google’s progress.
How Will Artificial Intelligence Change Search Engines in the Future?
For users, advances in artificial intelligence mean that finding the information they need will become much easier, particularly when they have previously searched for related information.
By combining AI with other technologies, search engines will become more efficient and provide more user-friendly interfaces.
For marketers, AI algorithms are making content marketing increasingly dominant. This means they need to shift their attention away from individual keywords and keyword phrases toward content that reflects users’ backgrounds, interests and intentions.
As online businesses become increasingly focused on their websites, attracting genuinely interested potential customers may require a longer-term approach.
Although the full impact of artificial intelligence is not yet understood, it is already playing a role in the algorithms that determine search rankings.
For marketers familiar with SEO, the traditional emphasis on keywords is declining. They urgently need to adopt new techniques to develop effective marketing strategies.
How Does Artificial Intelligence Affect Language?
Neural language processing and machine learning are being widely applied to create better translation systems.
Although numerous online translation programs are available today, most still contain grammatical errors and rely on relatively primitive data structures.
However, Google developed a Neural Machine Translation (NMT) system that is fundamentally different from its earlier translation technologies.
I tested the system and found that it still produced grammatical errors. However, it gradually learns from these mistakes and improves its performance. Whenever I have time, I help correct some of its grammatical errors.
Nevertheless, it is still too early to rely entirely on neural machine translation systems.
With the integration of artificial intelligence, these systems will gradually learn correct grammar and pronunciation over time. This is largely because their neural architectures are developed based on principles inspired by the biological human brain.
What Is Your Opinion on Search Engines and Big Data?
Search engine providers already offer statistical analysis capabilities.
Some search engine providers collect long-term data from every interaction through various methods and process this information using big data technologies.
They have developed big data processing frameworks that are compatible with search engines. These frameworks are scalable and capable of processing enormous volumes of content.
What Are the Current Trends in Search Engine Technology?
Real-time analysis and real-time search are essential components of modern search engines. The emergence of cloud computing has made implementing these capabilities much easier.
Cloud computing is transforming the world of computing in profound ways.
The current search engine trends can primarily be divided into three areas:
Big data
Real-time search personalization
Machine learning
Real-time personalization, for example, enables companies to improve the customer experience.
What Do You Think About Today’s Search Engines? Are They Perfect?
Search engines can understand a very broad range of general information. This is why many search engine providers continuously update their systems using different algorithms.
For example, Google Search can understand requests relating to films, maps and photographs. However, these are not necessarily the results users are actually looking for.
Users want more detailed and relevant information, something Google Search cannot always provide.
Every company has its own terminology and data. When companies conduct searches, they may receive very different results.
For example, if a cybersecurity company develops a specialized search application for the security industry, or a software company creates its own dedicated search engine, users may still be unable to find the information they need through conventional search technologies.
This is because most existing information is presented in broad, generalized formats.
Search engines can create patterns that match individual queries. They can use big data to identify and develop these patterns.
This means that search engines should build databases of patterns capable of matching users’ queries with the most relevant information.
However, these capabilities are still missing from many of today’s search engines and their underlying technologies.
Search engine providers need to develop a much deeper understanding of their users.
We need a search engine that we can genuinely understand and interact with.
This is why Google’s personalized search capabilities are still relatively primitive.
A personalized AI search system, however, could eventually eliminate the need to type search queries altogether. Instead, users could communicate their needs in a much more natural and sophisticated way, receiving results through deeper and more intelligent interactions.
Of course, there is also a disadvantage. If users have never previously searched for related information, they may struggle to find what they need. Since the search engine has very little available information to use as a reference, its final results may not be fully optimized.
