- Nowadays most of the tech firms are riding a wave of commercialization and pushing to create advanced Artificial Intelligence(AI) solutions.
- Most of the time, these firms are losing vision regarding the core problems to be solved or are creating problems to justify investing heavily in AI development.
- In the long run, experts believe such a race to build an ultimate AI model can have a harmful impact as well and be a threat to general security.
The shift toward Artificial Intelligence(AI) has gained an unprecedented speed in the recent past. More and more technological firms are pushing to be independent of Human Resources and shifting to a robot-mediated and controlled experience. There are many reasons for this increased dependence on AI applications. Whether it be the speed of offered solutions or the computational accuracy, AI offers a more fuss-free existence than humans could ever imagine. But now experts are wondering if this hype around AI has managed to overshadow the most crucial aspect of technology, solving problems at the core level.
A race to become “AI First”
In a bid to keep up with the AI revolution, firms are trying tooth and nail to be presented as “AI First”. Who can deny how many times words like “Artificial Intelligence”, “machine learning” and “deep learning” appear on various technology innovation associated platforms? But the real question which remains is whether such a radical introduction of AI in a consumer’s life is actually backed by relevant scientific research and backing or is it just a bid to confuse them and derive more profit out of investors and regulators.
In his new book named Doing AI, Richard Heimann, Chief AI Officer at Cybraics, answers these questions. Heimann’s take-home message remains that we tend to erect AI itself as our goal, we lose sight of all the important core problems we need solutions to. And thus in due course, we draw the wrong conclusions and make the wrong decisions. In simple terms, he believes that for tech innovators looking to become “AI First”, they need a lesson on how to do AI “Last”.
The need of the hour: To be solution-driven
The main topic of argument against the race to integrate AI lies in the shift of focus from the actual core task at hand, that is providing solutions. In a converse aspect, when firms put a large sum of capital into developing the latest and ultimate AI solutions, they tend to in turn “create” the problems such models are meant to solve in justification.
A prime example is a trend surrounding large language models, which are assuming a lot of hype in mainstream media and are being presented as general problem-solving solutions in natural language processing. While these models are in the real sense of the term “impressive”, however, they are not a silver bullet. In fact, in many cases, when firms have a well-defined problem, a simpler model or even a regular expression or rule-based program can be more reliable than advanced options such as GPT-3.
Heimann writes in his book that many times he feels as if tech firms try to advertise their AI applications in words via mind-numbing jargon that is way beyond understanding for the common man. Furthermore, upon dissection under trained eyes such as his, he finds that the actual problem being solved by said AI, appears to have been manipulated into being one or rather is not a problem at all. Heimann believes such firms are headed by wannabe- entrepreneurs who don’t love their business and are rather looking for a profitable acquisition under the wing of well-established firms such as Google.
A blur between the boundaries
Conventionally, there are two forms of research on Artificial Intelligence(AI). There is an academic AI branch that is headed by scientists and researchers who extensively study cognition, integrations in the brain, and behavior in both animals and humans to find hints about creating artificial intelligence. Hence, this branch is more dedicated to pushing the boundaries on what is already known about AI.
Applied AI, the other branch, on the polar end of the spectrum, aims to conduct research only to solve specific problems and ship products to the market. Developers of applied AI systems work for commercial purposes, that is to provide solutions to meet memory and computational constraints. Hence, this branch is more dedicated to products and profits and in general, the commercialization and exploitation of the knowledge on AI.
The AI hype has attracted considerable funding to the field of applied AI, providing startups and research labs with plenty of money. But it has also had its adverse effects. In recent years, riding on this wave of commercialization, companies like Google, Facebook, and Microsoft account for much of the money that goes into AI research. Consequently, their commercial goals affect the directions that AI research takes. As a result the line between academic and applied AI stands blurred. Also, using the ambiguous and vaguely defined term “AI” sets high expectations in clients and end-users and causes confusion. It can also drive companies into overlooking more affordable solutions and waste resources on unnecessary technology.
A question of safety
As already debated, the definition of AI is still not complete. While scientists are pushing their limits to develop the latest and most advanced technological models on this aspect, an important byproduct that people are turning a blind eye to is the aspect of safety.
Most of the AI being used in this era has been designed mindfully to make them more efficient than humans in terms of speed and accuracy but at the same time, to trigger “humanoid” responses. In simple terms, AI is built in a way that it can resemble human interactions. This was mainly accepted as a universal standard for the ease of integrating such technology into our daily lives. However, now it’s highly debatable if in a race to build the ultimate “AI” scientists will end up creating something undesirable.
