Exploring AI: The Past, Present and Future of ChatGPT

The Past, Present and Future of ChatGPT

Many have termed the late-2022 introduction of ChatGPT an "iPhone moment." But even the iPhone didn't break from the gates as fast as ChatGPT.¹ It attracted 100 million users within two months — 15 times faster than Instagram and 35 times faster than Uber.²

This potentially transformative technology has garnered seemingly limitless hype and makes the future of ChatGPT feel extremely bright. But global tech industry consultant Gartner notes that a new technology typically passes through a “hype cycle.” Inflated initial expectations lead to disillusionment before the technology matures to achieve widespread adoption.³

The computing power and intellectual framework underlying ChatGPT likely represent the next phase of decades-long development. This evolution has led to hardware and software capable of processing massive amounts of data and "learning" through repeated interaction.

Companies harnessing ChatGPT's potential could transform their businesses, products and services, creating new industries, jobs and opportunities.

But ChatGPT also has drawbacks. Get-rich schemes involving the technology have popped up across social media.⁴ There are legitimate concerns about job displacement. It has also drawn criticism about intellectual property rights. And ChatGPT still needs to be reliably accurate.

Some have called for regulation. Sam Altman told Congress that the "regulation of AI is essential" to limit its harmful impact on humans.⁵ Altman should know: He's the founder and CEO of OpenAI, the company that developed ChatGPT.

Dystopian sirens have sounded but reflect nebulous fears of an unknown future rather than current reality.

Generative AI remains in its infancy. What's more, there is no “intelligence” in artificial intelligence — it can’t think. It uses algorithms that predictive analytics have employed for the past two decades.⁶ So, it's more evolutionary than revolutionary.

What Is Generative Artificial Intelligence?

Open AI developed ChatGPT and DALL-E-2, an AI system that creates images via natural language processing.⁷ In March 2023 — just 3.5 months after ChatGPT's first public unveiling — the firm released ChatGPT-4, its most powerful version.⁸

"GPT" stands for "generative pre-trained transformer." It's a neural network modeled on the human brain with artificial neurons that connect to other artificial neurons and interact based on electrical stimuli. Collectively, this is called a "large language model." Large, in this sense, means a few hundred billion connections. By comparison, the human brain has more than 100 trillion connections.

Large language models, such as ChatGPT, are a type of machine learning that uses algorithms and statistical models to analyze data. For years, algorithms have analyzed numerical data to identify patterns and assess probabilities. Large language models can now do the same with text and visual images.

The "transformer" part of ChatGPT differentiates this new type of neural network from older versions, which formed the foundations for applications such as Apple’s Siri and Amazon’s Alexa virtual assistants. Google developed the first transformers in 2017.

The combination of all these — large language models, neural networks, machine learning, and transformers — form the basis for ChatGPT. The early focus for ChatGPT centered on its ability to aggregate and train based on 300 billion words in content on the World Wide Web.

From that review, it produces reams of readable text on specific topics. But the broader technology could apply to other domains. For instance, why not train it on the relationships between genes and cells?

What Can ChatGPT Do?

ChatGPT's main advantage is that it can absorb in a single day of training what it would take a human 150 years to learn. That should benefit a world that now generates more data every two days than it created from 10,000 B.C. through the 20^th century.⁹ By 2025, the world's data is expected to increase fivefold from just five years earlier.¹⁰

The technology's sheer power quickly attracted corporate America’s attention. Fifty-three percent of corporate information technology leaders plan to evaluate, use or allocate further resources to OpenAI for ChatGPT technology. That's a record for any single technology provider.¹¹

Consulting giant PwC expects AI to contribute positively to sales and earnings in every sector of the economy by 2030.¹²

Led by ChatGPT, those potential contributions seem endless.

In 2023, 27% of polled professionals said they had used ChatGPT to help complete work tasks.¹³ Experiments have shown that workers tackling short written tasks can complete them 37% faster if they use ChatGPT.¹⁴

In this way, ChatGPT helps humans improve work or content, but it can't do it alone — similar to how a soft drink needs a refrigerator to stay cold. But this enabling technology can help workers and industries exponentially by serving as an interface to any piece of software, boosting its capability.

What ChatGPT Is Doing Across Sectors

These uses barely hint at generative AI's potential. Not only will new and unforeseen uses emerge over time, but as its training improves and more powerful models surface, its capabilities will expand even further.

ChatGPT Has Limitations and Challenges

Despite its promise, we must understand ChatGPT's limitations, which may dramatically affect how it can be implemented in the future.

For all its comparisons to Google Search, it's not a search engine. It predicts the most statistically probable next word in a sentence based on its previous training and surrounding words.¹⁸ Furthermore, it needs humans to update it with new information continuously and requires feedback loops so that it can learn from its mistakes.

In the meantime, humans must provide guardrails. For instance, although ChatGPT can help doctors manage patients' inquiries, medical teams stop short of allowing it to provide specific medical advice.¹⁹

Potential for Disinformation Produced by Generative AI May Pose a Threat

ChatGPT's responses can’t provide a linked source to validate their accuracy. Perhaps most alarming, at times, it will respond inaccurately — sometimes wildly so, “hallucinating” with information it deems plausible. There are also cybersecurity concerns about the potential for exploitation by bad actors.

As a result, critics worry the technology will contribute dramatically to the "fake news" phenomenon. In early 2023, ChatGPT-authored articles in Men's Journal contained numerous factual errors.²⁰ A fake news story produced by ChatGPT and falsely attributed to The Guardian surfaced on the web.²¹ And a fabricated AI-generated image of Pope Francis wearing a fashionable white puffer jacket went viral.²²

Nonetheless, AI-generated news sites more than doubled to 125 in just two weeks during May 2023.²³ Media magnate Barry Diller said publishers should sue ChatGPT and other generative AI firms for copyright infringement. Indeed, a group of individuals seeking class-action status filed suit against OpenAI, claiming the company violated privacy laws by scraping 300 billion words from the internet to train its models.²⁴ A dispute over AI-created material also contributed to a 2023 strike by Hollywood screenwriters.

"We don't want our material feeding them," one screenwriter said of generative AI models, "and we also don't want to be fixing their sloppy first drafts."²⁵

Will Generative AI Replace White-Collar Jobs?

Writers and other written content producers lie at the heart of another widespread ChatGPT concern: job displacement.

Traditionally, technological innovation has threatened the blue-collar workforce. But ChatGPT seems to have set its sights on white-collar roles.

OpenAI and University of Pennsylvania researchers found that dishwashers, motorcycle mechanics and short-order cooks face no threat from generative AI replacing them. But they warned that journalists, mathematicians, interpreters and web designers face potential obsolescence.

Office and administrative support positions — secretaries, bill collectors, and office assistants — already employ 1.3 million fewer workers than in 1990. The U.S. Labor Department forecasts a further decline of 880,000 by 2031.

Past predictions of technology's detrimental impact on workers have often proved overly pessimistic. Take truck driving: A decade ago, economists predicted self-driving trucks would erase the need for human drivers. Today, U.S. roads have more human truck drivers than ever, and demand for them remains high.²⁶

What’s the Future of Investing in AI?

We believe that generative AI could present many potential investment opportunities. In the next part of our AI series, we examine those potential opportunities more thoroughly.

At a high level, simply training AI models still requires considerable computing power. Its nature and magnitude require so-called "parallel processing," with thousands of computer chips working side by side to run a single GPT model.

That's why Nvidia, whose stock has soared in 2024, is a popular AI investment target. The company specializes in high-end graphics processing chips that enable generative AI technology.

Cloud-based AI services may offer considerable advantages in productivity, efficiency and security. Amazon recently launched a model with 20 billion parameters for its Amazon Web Services customers. For its part, Google arguably has the most data of anyone on which to train its models.

Both companies have the technical expertise and computing power to create powerful tools for their customers. The distinction between them may rest on operational superiority rather than technical differentiation.

Finally, ChatGPT can’t currently reason or "think" logically. Some view that possibility as a threat; others see it as an opportunity. Either way, the technology will continue evolving, and developers will integrate new tools and functionality into the model.

As a result, ChatGPT will one day dramatically supersede its current capabilities with far fewer flaws.