Introducing OpenAI’s “Deep Search”
OpenAI’s “Deep Search” AI tool is one of the latest innovations in the realm of artificial intelligence, garnering significant attention in tech circles. This tool is distinguished by its ability to conduct deep analysis of information within minutes, saving experts hours that would otherwise be spent completing the same tasks.
Efficiency and Versatility
The technology behind Deep Search accelerates the processes of research, extraction, and interpretation of data efficiently. It can be utilized in various fields such as academic research, commercial data analysis, and technological development. This innovation reflects the rapid advancement in AI and its capability to enhance productivity while reducing reliance on human expertise for certain tasks.
Marketed as a research assistant capable of rivaling a trained analyst, Deep Search performs automatic internet searches, gathers sources, and presents organized reports. According to ZDNet, it achieved a score of 26.6% on the recent Human-Likeness Evaluation (HLE) test, a challenging benchmark for AI, outperforming many other models.
Advanced Technology with Drawbacks
Despite its polished outputs, Deep Search has its share of shortcomings. According to Science Alert, the tool does not entirely live up to the hype surrounding it. Experts have noted that it can miss essential details, struggle with handling recent information, and sometimes fabricate facts.
The company acknowledges that Deep Search “may occasionally invent facts in responses or draw incorrect conclusions, although this occurs at a significantly lower rate compared to current ChatGPT models, according to internal assessments.”
The Verge comments that it’s not surprising for unreliable data to slip through, as AI models do not “know” things in the same way humans do.
Questions and Concerns
The concept of an AI “research analyst” also raises several questions. Can a machine, regardless of its power, truly replace a human expert? What are the implications for knowledge work? Does AI genuinely help us think better, or does it merely facilitate our ceasing to think altogether?
In the numerous domains where generative AI has been tested, law may be the most prominent field displaying evident failure. Tools like ChatGPT have led to lawyers facing sanctions and experts encountering public embarrassment by producing documents based on fabricated cases and non-existent research citations.
The Balance Between AI and Human Expertise
Casey Newton from Platformer notes that like many AI tools, Deep Search works best if you are already familiar with the subject, partly because you can recognize where errors occur.
Just as the internet significantly lowered the cost of information transfer, AI is reducing the cost of cognition, according to Harvard Business School professor Karim Lakhani, who has studied AI and machine learning in the workplace for years.
With the public increasingly expecting companies to offer seamless and AI-enhanced experiences and transactions, Lakhani asserts that we must embrace this technology, learn how to leverage its potential and develop its use cases in our businesses.
Lakhani’s main point is that AI will not replace humans, “but humans working with AI will replace humans who do not work with AI.”
This is undoubtedly the case for generative AI. The first step is to begin: start experiments, create testing environments, organize training sessions for everyone to access the tools, identify the scenarios they will develop, and then implement them.
A Double-Edged Sword
In our increasingly digital world, AI has become an integral part of our daily lives, influencing how we shop, communicate, learn, and now, even how we work.
The intriguing debate over AI potentially replacing human jobs in the workplace generates varied opinions and concerns. With the changes already underway and the rapid pace of technological development, understanding the risks and opportunities is more crucial than ever.
The growing presence of AI in the workplace can be a double-edged sword. On one hand, AI offers opportunities for increased efficiency and the emergence of new jobs. On the other hand, it raises fears of job displacement and the loss of the human touch.
New Horizon states that AI is evolving so rapidly that it is understandable for many people to be concerned about job loss. When you pause to think about it, the implications are vast.
AI has already impacted jobs characterized by repetitive responsibilities and process dependence significantly. Roles such as translators, copywriters, graphic designers, and programmers are more susceptible to being replaced by AI.
These changes may seem threatening, and there will undoubtedly be disruption. However, this shift away from human roles may take many years.
While AI will transform the nature of work, human expertise, creativity, and emotional intelligence will remain essential in many fields. Many experts believe that by focusing on developing skills that complement AI, workers can adapt and thrive in the changing job market.
Despite AI’s “superhuman” capabilities, humans possess an intuition that AI has yet to replicate. Without intuition, AI can produce results that may appear correct but require more precise interpretation, something AI cannot achieve.
What Will AI Replace?
New Horizon suggests that AI will replace repetitive and routine tasks that require minimal creativity or emotional input, such as data entry, customer service chatbots, and some aspects of content creation.
However, AI will not replace human jobs; it will enhance existing roles, allowing companies to rely more on human capabilities.
By taking over routine tasks, AI will free up individuals to focus on the complex, strategic, and creative aspects of their roles that AI cannot handle.
Simultaneously, AI will augment human skills, enhancing workforce performance and efficiency. This symbiotic relationship between humans and AI will be vital for driving innovation and productivity in the future, with new technologies, marketing campaigns, and almost every industry relying on critical elements that only the human mind can provide.
The Role and Limitations of Deep Search
Marketed towards professionals in finance, science, policy, law, engineering, academics, journalists, and business strategists, Deep Search handles the heavy lifting and research within minutes.
Currently available only to ChatGPT Pro users in the United States at a cost of $200 per month, Deep Search follows a multi-step process to produce an organized report, unlike typical chatbots that provide quick answers. At first glance, it appears to be an ideal tool, but a closer look reveals significant limitations and flaws, such as summarizing a lot of information without fully understanding its importance.
Additionally, AI overlooks new developments, fails to account for critical legal rulings and scientific updates, and can generate incorrect information with confidence. It cannot distinguish between fact and fiction or discern reliable from unreliable sources.
Therefore, the summaries generated by AI do not match the depth of a skilled human researcher. No matter how fast AI is, it remains a tool, not a replacement for human intelligence.
The Need for Human Review
For knowledge workers, it has become more crucial than ever to invest in skills that AI cannot replicate, such as critical thinking, fact-checking, deep expertise, and creativity.
AI can be used judiciously to enhance research without sacrificing accuracy or depth. It can achieve efficiency, like summarizing documents, but ultimately, human judgment and expertise must prevail when making decisions.
Despite widespread marketing efforts trying to convince us otherwise, generative AI still has many limitations. Humans who can creatively integrate information, challenge assumptions, and think critically will remain in demand—AI cannot replace them yet.
Andrew Rogoyski, director of the Human-Centered AI Institute at the University of Surrey, warns of the risk that humans may use outputs from tools like Deep Search as they are without conducting retrospective checks on what the AI produces.
Rogoyski told The Guardian, “There is a fundamental problem with knowledge-based AI, which is that it would take humans many hours and a lot of work to verify whether the machine’s analysis is good.”
AI companies do not hide this issue. OpenAI, for instance, mentioned at the launch of Deep Search that the tool can “hallucinate” facts in answers or draw incorrect conclusions. They added that Deep Search users might struggle to differentiate between reliable information and rumors and may fail to convey uncertainty correctly, highlighting the need for human review.
