The computer program that talks back is taking over.

Netflix built a subscriber base of 1 million users in a mere 3.5 years following its launch, says statistics portal Statista. Twitter scored its first million tweeters in a snappy two years, Facebook its first million friends in a faster 10 months.

ChatGPT, the chatbot program that debuted near the close of 2021, landed its first million users within five days after its launch. In the two months following its launch, its subscriber base ballooned to more than 100 million users.

The bot’s runaway success is earned, say tech industry watchers. “The hype is warranted,” says Kemal Kvakic, Raymond James IT head of innovation.

What’s a chatbot?

A traditional chatbot is a computer program you can have a conversation with, through text messages or voice interactions. And for more than a decade they’ve filled commercial roles in customer service, mimicking human conversation, most often to answer a question you’ve asked.

Traditional chatbots depend upon “intent recognition.” The bot’s developers try to predict what you’ll ask and then program the bot with appropriate responses. If the bot can’t answer your question from what it’s been taught, it issues the all-purpose reply, “I’m sorry, but I don’t understand.” Congratulations – you’ve been chatbotted.

So, how does ChatGPT differ?
ChatGPT ups the ante.

Unlike traditional chatbots, ChatGPT’s doesn’t make simple matches between questions and preprogrammed answers. It digests vast quantities of data from across the internet, summarizes it, organizes it and taps into it to provide tailored answers just as a person would. And unlike other technologies, you can tell ChatGPT, “I didn’t understand that. Can you explain it another way?” And it will generate an updated response – one reflecting the entirety of your interaction with it so far.

It can even assume different identities. You can ask ChatGPT to explain something to you as if you’re an architect, a teacher or an 8-year-old, and it’ll respond accordingly. It can write computer code, poetry, lyrics and plays. It can generate images, audio and video. It’s so powerful, “The engineers who built ChatGPT don’t always understand why it provides the answers it does,” Kvakic says. And businesses have taken note.

Microsoft’s Bing search portal has already begun using ChatGPT to help enrich its platform. Microsoft’s total investment in OpenAI, the company behind ChatGPT, has reportedly reached $13 billion, and the company is leveraging OpenAI technology in Copilot, an AI-based assistant feature for Microsoft 365 apps.

Competition will surely grow in this space, says Kvakic, with more products integrating personal assistant-like capabilities quickly. Not to be one-upped by the likes of Bing, Google already has released its own advanced-AI chatbot, dubbed Bard.

In countless professions, the variety of generative AI driving ChatGPT could massively streamline associate training, customer service, code debugging and more. Imagine how this technology could create efficiencies – from cancer researchers who need to synthesize mountains of scientific data to legal teams who need to rationalize decades of case history.

With great power comes worry

For all its potential, ChatGPT and its emerging competitors pose risks. Concerns about information reliability and bias rest at the forefront of watchdogs’ worries, along with questions about the potential for deceptive deepfakes, copyright implications, privacy breaches and other uses by bad actors.

Not surprisingly, policy guardrails are pending. Several countries have enacted outright bans on certain types of AI systems, while U.S. policymakers are seeking public comment toward establishing rules to govern advanced AI systems in the states.

“We know AI regulation is coming,” says Kvakic. “The question is, how much?”

Facts and stats

  • AI and chatbot technology have been a work in progress for nearly 60 years.
  • The “GPT” in ChatGPT stands for “generative pre-trained transformer,” an AI technology developed by OpenAI.
  • Early leaders of OpenAI included Sam Altman, Elon Musk, Reid Hoffman and Jessica Livingston.
  • Microsoft holds a reported 49% stake in OpenAI.

Limitations and Unknowns

  • Cybersecurity. Fraudsters may be able to use the tool to write convincing phishing messages.
  • Objectivity. OpenAI’s CEO has admitted that ChatGPT has shortcomings around bias.
  • Regulation. AI regulation has already started to take shape, particularly in Europe.
  • Competition. It’s likely that competing tools will continue to emerge in the months ahead.