Will AI thrive or disappoint? Examining its breakthroughs, challenges, and what’s needed for long-term success.
It’s no secret that prominent tech predictions have proven to be inaccurate in the past; Bill Gates predicting spam wouldn’t exist by 2004, Clifford Stall claiming that cyber business was impractical, and Robert Metcalfe predicting that the internet would collapse – and now perhaps the AI singularity (AGI) by 2045.
From digital assistants and individualized learning to predicting the chances of breast cancer and fraud, AI has made a profound impact. Finance, healthcare, education and many other industries have adopted the technology in some way or the other and experienced positive changes.
Yet, issues are creeping up, deflating the rising technology’s image and potential. While the AI singularity may be a stretch, critics are also skeptical about AI’s current potential and how much value it is adding.
Some say the flames will only spread from here onwards, while others think the fire will eventually be extinguished.
How AI is on Fire
AI has taken the world by storm, and in only a few years, the technology has demonstrated its transformative potential–bringing disruptive innovation across so many different industries. In the past year, AI’s progress has been more incremental, focusing on refining capabilities, improving ethical frameworks, and setting stronger foundations for sustained impact.
In asset management, AI has been able to help power revenue growth by analyzing data for forecasting and eliminating human error in project management, saving businesses valuable time and resources. It also assists in several other processes, enhancing quality and accentuating human intelligence.
In healthcare, AI accelerates drug discovery, detects diseases early, and strengthens cybersecurity through advanced medical coding. In finance, AI enables traders to make split-second trading decisions, improves fraud detection, enhances customer service, and delivers powerful predictive analytics.
Generative AI has also transformed content creation, supporting everything from research to crafting targeted email campaigns.
Among recent milestones, face recognition technology has enabled the opening of the world’s first passport-free airport in Singapore, advancing border control automation. Additionally, the emergence of the world’s first AI travel influencer and the launch of SearchGPT as an alternative to traditional search engines underscore AI’s expanding role in reshaping user experiences and business processes.
AI’s rapid adoption has created a sense of urgency across industries. The technology’s flames are becoming so widespread that any business that isn’t adding AI features has the fear of missing out or becoming irrelevant. However, that fear may have no basis.
While AI’s economic potential is forecasted to reach $15.7 trillion by 2030, there is lingering skepticism about whether the technology will deliver on its projected value. Will these trillions come to fruition, or will AI fall short of expectations?
As AI reshapes industries and enhances productivity, the technology continues to fuel debates around its long-term value and impact, leaving room for reflection on the true scope of its promise.
The Spark That May Die Out
In Thinking Fast and Slow, Daniel Kahneman describes two systems of human thinking: System 1, fast and intuitive, and System 2, slow and analytical. Arguably, AI can mimic both systems and is free from human limitations such as preferring cognitive ease, heuristics causing biases or prejudice, and having trouble reconstructing past situations.
However, AI lacks contextual understanding, creativity, emotional intelligence, and logical reasoning, which means that it may outperform human intuition (System 1) in some areas, but falls short of the nuanced analytical depth of System 2.
Moreover, AI is vulnerable to its own biases, known as “AI bias,” driven by data inaccuracies inherent in training data. A significant flaw in generative AI systems is their tendency to hallucinate, producing responses based on nonexistent data, which is a serious reliability issue that is eroding user trust.
A notable case involved a New York lawyer relying on ChatGPT, which fabricated six legal precedents—one of the early detections of the technology’s ability to hallucinate. GPT-4o is currently 88.7% accurate while GPT 3.5 (and older models) have higher probabilities of hallucinations and inaccuracy.
Google’s Gemini has shown improvements in accuracy, but no AI tool is 100% reliable. Studies suggest that GPT is likely to become less accurate with time—a process known as a “model drift”.
There are several ethical concerns with the use of AI that are leading to intense debate in all industries. The concerns center on data privacy, its role in creating potential harmful deepfakes that can help administer criminal activity, and the technology’s influence in fields where creativity and originality are paramount. AI has played a role in developing deepfakes that have influenced the US election.
Additionally, an increasing number of AI projects are failing to produce return on investment. A study by Harvard Business School stated that about 80% of industrial AI projects may not realize tangible returns and this can severely discourage investment in the technology and hinder progress.
An example of this is the “productivity paradox,” where despite heavy investments, AI’s promised productivity gains are also not as visible in the tech sector, leading to questions about its economic benefits. According to SaaSLetter’s “State of AI” report, productivity gains lie between 5-20%. AI users also report moderate quality improvements averaging between 5 and 20%. Though this is better than nothing, it still falls short of the massive expectations the technology has generated.
Although AI continues to grow, these limitations suggest that the current enthusiasm may wane if challenges are not addressed.
Fizzle Now and Sizzle Later
Looking at the current state of AI, the technology’s hype and the enthusiasm surrounding it may die down temporarily—-but with gradual improvements, AI is set to sizzle and take the world by storm. Here are a few recommendations that would make the tech more sustainable and impactful.
- Redefine how AI value is measured: CJ Gustafson mentions in Mostly Metrics that the value generated from AI projects can’t be measured according to the conventional SaaS model. AI revenue is “reoccurring” , not “recurring” which means it is generated as and when the need arises (most AI tools charge per output).
It’s important to reassess how to measure the success of AI projects as often, the benefit of these tools won’t be directly quantifiable. It’s important to educate business owners and institutions on how to measure the impact of AI on their business practices before the technology is termed a “failure” because of intangible returns. - Focus on better rather than faster: Productivity gains are important, but they can’t be realized until AI is accurate — otherwise, the time saved will be spent on validation. Therefore, it is important to enhance the quality and reliability of AI output rather than focus on its speed alone. Although the AGI is perhaps an unrealizable dream, we’ll only be close to it once output accuracy is improved.
Also, AI tools should be trained on contextual understanding and emotional intelligence, helping them become more apt at decision-making and eliminating aspects of AI bias. - Target high-impact use cases: Those using AI right now are still what can be termed as early adopters—-while the technology has caused excitement, it has been unable to meet expectations because they superseded reality.
Also, to make AI indispensable, it should be deployed to service paramount use cases or to solve essential problems. At the moment, it is a nice-to-have but not a must-have and to position it as such, it needs to help with unique challenges outside of human capability. For example, features such as Apple Intelligence are raising eyebrows, but aren’t intriguing or revolutionary enough to encourage consumer investment. The same issue prevails for many AI solutions. - Address ethical concerns and regulate use: It’s evident that it’s important to regulate AI use and address ethical concerns, such as the ability to create deepfakes. Industries such as the education sector are already adopting policies to regulate the use of AI and use the technology responsibly. Broader regulatory measures are necessary to address privacy, content authenticity, and misuses.
By focusing on meaningful metrics, quality, essential use cases, and ethical safeguards, AI’s diminishing flicker of potential can evolve into a sustained, powerful force across industries. These actions will ensure that AI remains both relevant and impactful, turning any skepticism into long-term success.
Explore AITechPark for the latest advancements in AI, IOT, Cybersecurity, AITech News, and insightful updates from industry experts!