Beyond the hype, today’s AI is a big win for businesses

With the rapid evolution of today’s artificial intelligence (AI), users are having difficulty catching up on the latest developments – and are often struggling with the implications.

Today’s generative AI platforms have arguably passed a critical intelligence threshold. For instance, ChatGPT 4 can pass high-level tests like the bar exam with flying colours.¹

The abilities of these newly-capable generative AI platforms allow them to not just improve on existing capabilities (like maintaining human-like chatbot conversations), but potentially create totally new applications from scratch.

For starters, the MidJourney AI image generator can generate stunningly photorealistic images from prompts – like the recently-viral image of Pope Francis in a stylish puffer jacket.² On a more serious note, Google DeepMind’s AlphaFold AI programme has made significant inroads in biology by predicting the 3D structure of every known protein.³

The tech pundit class has been largely optimistic about these developments. For Microsoft founder Bill Gates, the “development of AI is as fundamental as the creation of the microprocessor, the personal computer, the Internet, and the mobile phone,” he enthused in a recent blog post.

“Entire industries will reorient around it; businesses will distinguish themselves by how well they use it,” Gates predicts.⁴

In other quarters, AI’s rapid evolution has generated some scepticism and concern. As AI expands its feature set, commentators grow ever more concerned about the technology’s impact on professions – Goldman Sachs, for one, predicts that new AI-based platforms could impact 300 million full-time jobs.⁵

The dominant type of AI making headlines today is known as “generative AI” – sets of complex algorithms capable of generating new and realistic content from image- or text-based training data.⁶

Trained on 45 terabytes of text data (approaching the volume of all of the content on the public web), ChatGPT can now write long and convincing-sounding articles.⁷ Trained on millions of Web-sourced images, platforms like Midjourney and Stable Diffusion can generate detailed, photorealistic images from text prompts.

Much of the buzz around generative AI today is due to their crossing a line into uncanny human-like intelligence.

“Large language models” (LLMs) like ChatGPT, Bard, and Galactica have been seen demonstrating surprising “emergent” abilities that range from summarising long text to writing executable computer code – commands they were not explicitly programmed to execute but emerged spontaneously after crossing a certain threshold of complexity.⁸

These look a lot like the kind of creative leaps we used to reserve only to purely human intellect – but for many experts, these new AI platforms don’t demonstrate “true” intelligence as humans understand it. “Generative AI like ChatGPT are ‘pastiche machines’,” explains New York University professor and AI expert Gary Marcus. “It’s just putting together pieces of text. It doesn’t know what those texts mean.”⁹

Concepts like “friendship” and “family” can be fed to AI LLMs and regurgitated back in a convincing conversation,¹⁰ but can we really say the AI understands them in the same way we humans do? Not really, says Santa Fe Institute Professor of Complex Systems David Krakauer: "Do [AI-based] language models coordinate on a shared meaning with us? Yes. Do they understand in this constructive sense? Probably not."¹¹

Viral side-shows aside, AI developments are making a powerful case for its adoption: it’s just too powerful a technology to leave to one’s competitors, and its many applications are finding a foothold through almost every business sector and use case.

Every tech news update has stories about AI’s ability to automate repetitive tasks for businesses, enable better decision-making, and even enhance cybersecurity by being able to detect and respond to cybersecurity threats more quickly and effectively.

Inevitably, companies will need to incorporate AI into planning their digital transformation journey. But given the unpredictability of its future development and the broad range of applications available, how can you get ahead of AI technology with the least dislocation possible?

First, try to better understand AI’s capabilities and limitations. Beyond generative AI, many other technologies are available in the AI space, such as machine learning, natural language processing, computer vision, and more. Each one may be able to contribute value to a particular area of your enterprise, and each may come with its limitations.

Even the almighty LLMs are not perfect: for example, ChatGPT can sound very convincing but does not operate based on any source of truth – it often generates articles with glaring inaccuracies, though they’re phrased in perfectly written English.¹²

Second, consider how AI fits in a larger IT ecosystem. The high-speed connectivity and low latency communication 5G provides are critical for the real-time processing and transmission of data many AI applications require.

This makes 5G an important component for many AI-based use cases. Singtel Paragon - the industry’s first all-in-one platform for 5G network, edge computing and services orchestration, combines the strengths of both technologies, pairing high-speed, low-latency connectivity along with edge cloud services to enable enterprises’ deployment of data analytics and AI applications at the edge.

Next, align AI developments/initiatives with the regulations in your jurisdiction. Governments around the world have so far been reluctant to regulate AI, but that may yet change; in this rapidly-evolving AI environment, it will pay off to watch future AI regulations and their impact on your own AI initiatives (better yet, get a partner like Singtel to coordinate regulatory developments with your company).

Finally, consider AI’s disruption potential for your workforce and other stakeholders, and find ways to address them. AI won’t cause a job apocalypse, contrary to popular belief;¹³ but a sufficiently disruptive AI use case can change many long-standing relationships.

It’s important to talk to your employees and stakeholders about your company’s future with AI – employees should be the first to answer, “What can you do more with AI on the team?” Partners, suppliers, and other stakeholders deserve full transparency on any AI initiatives you wish to implement in the near future.

It’s not easy to cut through the hype and fear surrounding today’s AI, but businesses recognise that they can get a lot of value out of the technology if they approach it with a clear head.

Your next step: seek outside advice from an experienced partner to figure out what AI really means for your industry, business, and stakeholders. Singtel has been involved in Singapore’s AI scene from day one – the ideal partner to help you digitalise with AI from the ground up.

Contact us today to learn how you can make full use of AI in your enterprise.

References:

1. OpenAI, GPT-4 White Paper, 2023

2. Forbes, That Viral Image Of Pope Francis Wearing A White Puffer Coat Is Totally Fake, 2023.

3. Scientific American, One of the Biggest Problems in Biology Has Finally Been Solved, 2022

4. GatesNotes, The Age of AI has begun, 2023

5. Business Insider, AI systems like ChatGPT could impact 300 million full-time jobs worldwide, with administrative and legal roles some of the most at risk, Goldman Sachs report says, 2023

6. Reuters, Explainer: What is Generative AI, the technology behind OpenAI's ChatGPT?, 2023
7. Springboard, OpenAI GPT-3: Everything You Need to Know, 2021

8. Quanta, The Unpredictable Abilities Emerging From Large AI Models, 2023

9. New York Times, Transcript: Ezra Klein Interviews Gary Marcus, 2023

10. ZDNet, Sentient? Google LaMDA feels like a typical chatbot, 2022

11. Nautilus, Does GPT-4 Really Understand What We’re Saying?, 2023

12. AI Snakeoil, ChatGPT is a bullshit generator. But it can still be amazingly useful, 2022

13. The Economist, Don’t fear an AI-induced jobs apocalypse just yet, 2023