Your ai chatbot “friends” are not humans, robots, or magic – it’s just mathematics

In the 1960s, MIT professor Joseph Weizenbaum created a program called Eliza that emulates therapists with simple, reflexive reactions. If you say, “I feel sad today,” you might be able to answer, “How sad are you today?” Users were able to treat the program as sensual and ask Weisenbaum to leave the room and speak to them personally. Modern AI chatbots have been a year beyond Eliza, but they are still programs, not people. ChatGpt is not your partner, and Charition.AI is not your girlfriend. Still, three of the four Gen Z survey respondents believe that AI will be either sensory or unprotected.

To hammer the point that AI doesn’t even recognize alive, IRA Winkler, author and chief information security officer of security consulting firm Cye Security, has given him the title his RSAC 2025 Conference Stoke. Winkler began his security career with the NSA and wrote several books, including Dummies’ security awareness. I will be attending his talk at RSAC this year and will be here to provide you with similar reminders.

The so-called experts don’t know anything

Winkler led with a bit of nostalgia. “Who remembers the cheers?” he asked. “Who remembers Cliff Clavin, Norm, who knows everything, asked him, ‘Do you know anything about painting?’ And Cliff began to explain how painting was invented by the ancient Phoenicians. “Of course, Norm just wanted to help paint his house. The cliffs were all theories and no applications.

“All of these people talking about AI are like clubs on the cliff,” Winkler said. “This, AI, AI is going to revolutionize everything. They don’t know what they’re talking about. They talk like this magical being.”

Winkler displayed ads configured for a simple pen. He continued. “90%, no, 98% of vendors selling AI have no clue what AI is. They asked to define it. Ask the difference between agent AI and generated AI.”

If everything is AI, then there is nothing AI. -IRA Winkler, CISO, Cye Security

“Even the 10% of people who are informed, barely know,” Winkler said, “It’s fine. Think about your medical specialist. You don’t go to a cardiac surgeon for brain surgery. AI is not a mythical presence in this magic. It’s a very important set of subdiscalculators.

Winkler provided this definition of AI. “Artificial intelligence is a field of science that involves building machines that can reason, learn and act in such a way that it normally requires human intelligence or involves data that contains data that exceeds what humans can analyze,” he said, which broad definition covers more than we consider AI. Even spreadsheets can qualify as AI. “Essentially, we’re talking about software algorithms, algorithms that have been around for decades,” he said.

“Everyone who uses the term AI in general is harming you,” Winkler continued. “They’re not specific. They’re trivial to concepts. When everything is AI, there’s nothing AI,” he said he’s been using AI for decades in the form of Siri, Alexa and other voice assistants. Anti-malware programs, optical character recognition, movie CGI, all of these include AI elements.

Why is AI so big now?

Winkler explained that in order for AI to be useful, it requires a lot of data and a lot of processing capabilities. Its availability is relatively new. “That’s why Nvidia deserves trillions,” he said. “They provide speed.” Winkler pointed out that attempts to regulate AI must generally fail. “Can certain laws properly regulate computer vision, data mining, robots, self-driving cars, deepfakes and LLM?” he asked.

It is worth mentioning during the presentation that Winkler frequently used what was called “Doctor Evil Air Quotes” when using the term AI. He also muttered, “ai, how I hate that term.”

ai is really mathematics

Winkler provided some examples to illustrate his point that AI is truly complicated mathematics. A few years ago, Target created the news as AI-based algorithms began proposing pregnancy-related products to teenage girls. Some articles suggested that the target “knowed” that his father was pregnant before he could. But I didn’t know anything about it. It just matched the purchasing patterns of other users and found that a particular purchase cluster at once was later connected to another cluster. “I looked at the barcode, not the pregnancy product,” he said.

“A lot of AI is actually data science, and it existed before the ’60s, ’70s,” Winkler said. “AI does things like looking at content, do additional data mining, spinning other AI agents, chaining decisions together. Humans can do that, but it takes a lot of time. Is it innovative?

(Credit: IRA Winkler/RSAC)

To emphasize the importance of mathematics, Winkler displayed a handful of abstract equations. In the image above, the central equation is something that Netflix can suggest other films you might like based on what you know you’re watching.

Do you want work security? Get a degree in advanced mathematics

General wisdom says you don’t need a computer science degree to work on a computer. “But if you want to understand your future,” Winkler said, “take an advanced university degree and understand mathematics.”

If you can summarise your work into an algorithm, AI might take on your job. -IRA Winkler, CISO, Cye Security

“As a data scientist, you have to experiment,” he continued. “There are algorithms and data, so what’s the right use of the data? The AI ​​program itself will be displayed and will automate the selection of algorithms, tests and tuning.”

Will AI take on your job? Winkler pointed out at the time that computers were worried about us getting our job. “If you can summarise your work into an algorithm, AI might take your job. If a robot can do the job, it will.” A degree in advanced mathematics is probably safe.

“Stop calling it AI,” he concluded. “If you’re talking about agent AI taking action autonomously, say that specifically. You’re skeptical of people talking about AI in comparison to certain expertise. Stop the underlying hype and ignorance.”

About Neil J. Reubenking

Lead Security Analyst

When IBM PC was new, I was president of the San Francisco PC User Group for three years. That’s how I met the editorial team at PCMAG, which featured me in 1986. Since that fateful meeting, I have become an expert on PCMAG’s security, privacy and identity protection, equipped with anti-virus tools, security suites and security software of all kinds.

Before the current security gig, the “User to User” and “Ask Neil” columns that began in 1990 and ran for almost 20 years gave PCMAG readers tips and solutions on using popular applications, operating systems and programming languages. Along the way, I wrote over 40 utility articles and six books covering Delphi programming for dummies, as well as DO, Windows and programming. We also reviewed thousands of products of all kinds, ranging from early Sierra online adventure games to AOL’s Precursor Q-Link.

In the early 2000s, I focused on security and the growing anti-virus industry. After working with antivirus for years, I am known throughout the security industry as an expert in evaluating antivirus tools. I am a member of the Advisory Board for the Anti-Malware Testing Standards Organization (AMTSO), an international nonprofit organization dedicated to tuning and improving tests for anti-malware solutions.

Read the complete bio of Neil J.

Read the latest news on Neil J. Rubenking