In recent years, artificial intelligence (AI) has radically changed its shape, but will it ever be able to reach the level of human intelligence? The experts’ answer seems clear, writes Rozmova.

What initially caught the public’s attention as a growing field with some promising (but mostly uninteresting) applications has recently turned into a $100 billion+ industry, with big names like Microsoft, Google, and OpenAIs seem intent on outdoing each other.

The results have led to complex language models, sometimes released hastily and without the necessary testing and control.

These models can do what a human does, and in many cases they can do better at the tasks they are given. They can beat us in complex strategy games, create incredible art, diagnose serious diseases, and compose lyrics or music.

There is no doubt that AI systems look “intelligent” to some degree. But can they become as smart as us?

Experts have already coined a term for this: artificial general intelligence (AGI). While this is currently a broad term, it can be seen simply as the point at which AGI reaches the level of general human cognitive abilities.

Even more simply, this will be the point at which an artificial intelligence system can solve any intellectual task available to a human. Clearly, AGI has not reached that point yet, as current AI models are held back by the lack of certain human traits, such as true creativity or awareness of the emotions of the people they interact with.

Here’s what 5 experts say about the possible transformation of AI into AGI.

Consciousness, a problem for artificial intelligence?

Paul Formosa (artificial intelligence expert and philosophy of technology):

AI has already reached and surpassed human intelligence in many tasks. It can beat us at strategy games like Go, Chess, StarCraft, and Diplomacy, outperform us on many language learning tests, and write essays that would pass a university exam.

Of course, artificial intelligence can also make things up, “hallucinate” or simply be wrong. But so can people, albeit in different ways.

Given a fairly long time frame, it seems likely that AI will reach AGI, or “human-level intelligence.” This means that she has achieved a high level of knowledge in sufficiently interconnected areas of intelligence that people possess.

Some people still worry, despite the advances in artificial intelligence so far, that it will never become truly “intelligent” because it cannot understand what it is doing because it is not conscious. But the development of artificial intelligence suggests that we can have intelligence without consciousness, because intelligence can be understood in functional terms.

An intelligent entity can do intelligent things, such as learning, reasoning, writing essays, or using tools. The artificial intelligence we create may never have consciousness, but it is increasingly capable of doing intelligent things.

In some cases, they are already doing it at a level above ours, and this trend is likely to continue.

PHOTO: Paul Hartl / imageBROKER / Profimedia Images

What artificial intelligence systems are still lacking

Christina Mager (expert in computational neuroscience and biomedical engineering):

AI will reach human-level intelligence, but perhaps not very soon. Human intelligence allows us to reason, solve problems, and make decisions. It requires many cognitive skills, including adaptability, social intelligence, and experiential learning.

AI already meets many of these requirements. What remains a model of artificial intelligence is to study human-specific traits such as critical thinking, understanding emotions and the events that can cause them.

As humans, we are learning and going through it from the moment we are born. Our first experience of “happiness” is so early that we cannot even remember it.

We also learn to think critically and regulate emotions in childhood and develop a sense of “emotion” as we interact with and learn about the world around us. It is important to note that the human brain can take many years to develop this type of intelligence.

AI has not yet acquired these capabilities, but it probably will – and perhaps even sooner. We are still discovering how to build, train, and interact with AI models to develop such traits.

The really important question is not whether artificial intelligence will reach human level, but when and how.

What the recent past shows us about the evolution of artificial intelligence

Seyedali Mirjalili (artificial intelligence and swarm intelligence expert):

I believe that AI will surpass human intelligence. why the Past gives us some insights we cannot ignore.

Many people thought that tasks like playing computer games, recognizing images, and creating content (among other things) could only be done by humans. But technological progress proved otherwise.

The rapid development and implementation of artificial intelligence algorithms, along with the abundance of data and computing resources today, has led to a level of artificial intelligence and automation that was previously unimaginable. If we follow the same trajectory, having more generalized AI is not just a possibility, but a certainty in the future.

It’s only a matter of time. For example, artificial intelligence has come a long way, but not yet in tasks that require intuition, empathy, and creativity. But the development of algorithms will allow it to be done.

Moreover, once AI systems reach human cognitive abilities, we will see a “clotting effect” and AI systems will be able to improve themselves with little or no human input.

Such “intelligence automation” will fundamentally change the world. Achieving AGI remains a major challenge, and there are numerous ethical and societal implications that need to be approached with great caution as we continue to move toward it.

PHOTO: Cover image / Zuma Press / Profimedia Images

Quantum computers can unlock the potential of AI

Dana Rezazadegan (artificial intelligence and data science expert):

Yes, artificial intelligence will become as intelligent as humans in many ways, but how intelligent it will become will largely depend on advances in quantum computing.

Human intelligence is not just about knowing things. It has several aspects, such as creativity, emotional intelligence, and intuition, that current AI models can mimic but cannot match. That being said, artificial intelligence has come a long way and this trend will continue.

Current models are limited by the datasets available for training them, as well as limitations of quantum computing power. But the advent of quantum computing will change the capabilities of AI. With artificial intelligence augmented with quantum technology, we will be able to populate AI models with massive data sets comparable to the natural data set that humans acquire when interacting with the world around them.

These models will be able to perform and support fast and accurate analyses. The availability of an improved version of continuous learning should lead to highly complex artificial intelligence systems that, after a certain point, will be able to improve without human intervention.

Thus, artificial intelligence algorithms running on stable quantum computers will have a high probability of achieving something like generalized human intelligence – even if the artificial intelligence will not necessarily be on par with all aspects of human intelligence as we know it.

When can we expect AGI-type artificial intelligence

Marcel Schart (machine learning and AI leveling expert):

I think it is likely that AGI will become a reality one day, although the time frame in which this will happen remains very uncertain. If AGI is to be achieved, exceeding human intelligence seems inevitable.

Humans themselves are proof that the laws of physics allow for an adaptive and extremely flexible intelligence. There is no fundamental reason to believe that machines are fundamentally incapable of performing the computations required to achieve problem-solving skills such as those available to humans.

In addition, artificial intelligence has a number of distinct advantages over humans, such as better processing speed and greater memory capacity, fewer physical limitations, and the potential to be more rational and self-improving in an unlimited mode.

As computing power increases, artificial intelligence systems will eventually exceed the computing power of the human brain. Our main goal is to achieve a better understanding of intelligence itself and the knowledge associated with the creation of AGI.

Modern artificial intelligence systems have numerous limitations, and they are far from being able to master the various areas that characterize AGI. The road to AGI is likely to be littered with unforeseen advances and innovations.

Metaculus, an authoritative statistical forecasting platform, estimates the average date for AGI to be reached in 2023. This seems to me to be an overly optimistic forecast.

A survey of experts in 2022 found that there is a 50% chance that we will achieve artificial intelligence with human intelligence by 2059. This estimate seems more plausible to me.