Is AI technology really that hyped, or is it real? I started with this question and researched a bit, and this is the outcome of the research. Sit back and enjoy the ride.
“Before we work on artificial intelligence, why don’t we do something about natural stupidity?” – Jean Baudrillad (French Philosopher)
Recently, AI has become an integral part of many of our everyday tasks, even without our awareness. Both the trend and the capabilities are getting better as you read on. Many businesses and even some crucial government agencies use real-world examples daily. Furthermore, Hollywood presents an oversimplified portrayal of AI. So, you wonder: Is artificial intelligence a genuine concept?
Another perspective is that this technology has attracted a large investment. From venture capitalists and large corporations during the last five to six years. Is the next bubble, the artificial intelligence bubble? Is it about to burst, following in the footsteps of the dot-com boom? Will these massive, multi-billion-dollar investments yield returns? Does AI represent a wave or a bubble? At the present time, very few can provide an answer.
Flashback
From the Turing machine onwards, artificial intelligence has come a long way. All the AI startups are raking in cash for research and development in this current wave. The science fiction films made by Hollywood are completely at odds with reality. Current AI is only capable of doing boring, repetitive tasks. What the layperson would call “boredom work.” If Hollywood’s predictions were accurate by now, the scene would be different. We should see many armed, completely autonomous robots roaming the planet.
Most of today’s top robots can’t do much beyond answering basic questions. Like, “Alexa, what’s the weather today?” or performing document classification or predictive analysis. (You could glean it by gazing out the window.) Right now, those solutions function on social media sites. So, what gives rise to the disconnection?
Let’s assume a reporter were to hear that a Japanese robot solved the Rubik’s cube in under a minute. It is not considered to be breaking news. Actually, up until now, we have confined all the AI advancements to the category of “boring tasks” alone. To make any content lucrative, today’s news outlets’ inherent dimwitted reporters consume it. Artificial General Intelligence (AGI), humanoids, and talking robots are popular. Popular only in science fiction and popular cultures. Thus, it is necessary to spruce up the news content.
Where are we now in AI?
Can the current AI do all these? The short answer is “NO”. They are not even close to being able to do any of the above-mentioned tasks. Why? We have not succeeded in completely eliminating the human-in-loop in the system.
Is artificial intelligence real or hype? So now we’re back where we started.
Apart from this, the industry continues to brag about its ambitious AI-based promises. Let us also assume that these claims are true and treat them as such. Then I would like to know: Where are the self-driving cars? Where is the promised (in the early 2000s) next generation of intelligent servers? How come voice assistants still lack intelligence? This is a lengthy list.
The truth is that AI has been quite successful in certain niche markets. Many solutions have relied on augmented intelligence involving humans in some capacity. Back in the ’90s, Caterpillar introduced driverless, autonomous trucks to navigate mines. So, removing humans from the equation is essential for AI to reach its full potential. Then only it can generate the expected ROI.
The research on AI funding
I can run the numbers for you to see how much money corporations have invested in AI. To put it in context, early-stage funding for artificial intelligence and machine learning companies averaged around $4.8 million in 2010. But in 2017, the total funding for first-round early-stage investments reached $11.7 million, and in 2018, AI companies raised an all-time high of over $9.3 billion. As of this year, 2024 Perplexity AI has generated $136.3 million, and Celestia AI a whopping $175 million. These are apart from the lead OpenAI. (As reported by Forbes.com)
HPE acquired Cray, the industry leader in supercomputing, in May 2019. To aid in their foray into self-driving cars, Uber’s Advanced Technologies Group acquired Mighty AI in June 2019. Mighty AI specialises in developing training data for computer vision models. Microsoft purchased Nuance Communications in 2022 for a record $19 billion. It’s the highest price tag for any of the tech firms’ AI deals in healthcare. Last year, Amazon made an addition to its growing AI arsenal by acquiring Snackable.AI. It’s a small yet promising firm specialising in audio-focused artificial intelligence. The merger is to enrich its podcast offerings on Amazon Music. Among the many acquisitions that have been happening, these are but a handful. (Reference: Forbes.com, same as before.)
A pattern is beginning to take shape here, isn’t it? Correct. We find 2 things. “Software-based” AI companies are thriving. While physical manufacturing companies are still battling for survival, the truth is that physical models are very challenging. For software, performance is easy to develop and deliver.
The moot question on AI
This brings us to the crux of the matter: we view AI from a variety of angles. From chatbots to hyper personalised ads, the AI conversation covers it all, even in this blog. A layperson with a Hollywood lens finds these real-world solutions very intriguing. As an example, “Terminator” or “Westworld” come to mind.
A layperson’s understanding of AI is what’s known as AGI (Artificial General Intelligence). This refers to a supercomputer that mimics human thought and behaviour. In their pursuit of greater investment opportunities, companies are after building AGI. They have already declared their intention to construct an AGI model. Even if they cannot provide any assurance that their efforts will be successful. To acquire funding, they need to exaggerate the facts and their abilities. That’s where the hype begins and has persisted.
We have a long way to go before we can build this machine. If you recall, Sujatha wrote ” என் இனிய இயந்திரா” (My Dear Machine) in the 1980s, and we still haven’t finished even 10% of the story. Artificial intelligence programmers also wonder if we will ever get there. Most people dream of flying cars or talking robot servants, like in the movies. Yet these fantastical ideas rely on a critical factor. A factor called “AGI” is already innate to humans.
We should stop chasing after AGI and instead focus on a very specific area where AI is useful. Take a look at a few practical examples. Like chatbots for customer service, intelligent automation in mechanical productions. Even automated code corrections in SW development, and automated document classification. These are excellent instances where AI has proven to be beneficial. All it’s doing, in simple terms, is analysing past and present data.
Who needs another complicated system when we can do it all ourselves? Another serious question I’d like everyone to think about. Whether we should actually outsource our intelligence?
My findings
My investigation into this matter led me to discover that there are five key questions. When answered, it indeed reveals whether a proposed solution is intelligent.
1. Is there a trial version you can try out?
The vast majority of product showcasers would only ever show themselves. Because the solution is not yet stable enough for a public release. It’s tested in a confined setting before being considered a demo. If you were to request this, you would be familiar with the process.
2. What is the solution’s generality?
When we hear about a solution, like an AI reader, that mimics your reading experience, we take notice. Besides being attractive, artificial intelligence solutions, are very limited in their scope. The AI reader, in fact, can read the entire book faster than humans. It still can’t put itself in the shoes of a character, understand the subtle meanings in text. Why it cannot even answer a basic question like the story’s moral. As a result, there is never a broad solution.
3. Consider the implications of claims that it performs as well as or better than a human.
Artificial intelligence (AI) can automate repetitive tasks and execute them quicker than humans. But it lacks the artistic ability to perform at a human level. If the solution doesn’t show any signs of human creativity, then all they’re doing is automating it. Like building a statistical model to predict solutions from data they’ve already collected. It is devoid of any inherent intelligence.
4. Will it lead to basic AI finally?
Consider DeepMinds’s AlphaGo as an example. This is a traditional 19×19 board game with deep Chinese roots. Is it possible to use this solution in a 31×31 or 9×9 game with a human opponent? Then you might be aware that they need to prepare it. They need to retrain the system to play an extra 10 million games in the 31×31 format. Is it an intelligent solution if it cannot grasp the simple idea of the game?
5. What is its durability?
This solution caught my eye while I was looking into this subject. I couldn’t stop thinking about it until I had to ask this question. For blindness, their answer was an alerting device. The sensors will pinpoint the items, verify their identities, and signal back. It will alert the blind man to find the obstacles. Until I read the review at the end of the topic, I thought it looked great. Yet, it doesn’t work well in low light, bright midday, slanted sunlight, etc. Thus, while the solution is doable, it is not robust enough to withstand all types of weather.
Finally, I’d like to mention that current AI does rely on machine learning. It’s confined to rigorous mathematics, heavy statistics, and a limited problem set. The question your wife may ask, “Am I looking fat in this dress?” is not something that artificial intelligence can answer.
Take a shot with ChatGPT or Gemini and get back to me with the result. Even a horrendous husband would be able to answer this question at any moment.
I rest my case.
Very nicely articulated Vasu! AI is very much real and will only get more real as the cloud and computing costs get more affordable. It will take away a lot of mundane jobs – especially anyone who is manipulating spreadsheets – including project managers!