Martin Helberg talks about artificial intelligence

In less than two years, it has become public and widely used OpenAI’s ChatGPT chatbot and artificial intelligence have reached amazing milestonesmarking the before and after in various fields of knowledge and the way we interact with technology on a daily basis.

The developments have been so significant that academics consider 2023 a historic milestone with the development of several tools powered by generative AI, such as DALL-E 2’s image generator, Github’s Copilot, MidJourney’s art generator, and DeepMind’s AlphaCode. The programming problem-solving system, as well as new versions of ChatGPT, Bing, and Gemini, which use advanced AI algorithms to generate new content, whether in text, images, code, or videos, from descriptions or short instructions (prompts).

What is the common denominator between all these platforms, according to the professor Martin Hilbert, University of California, is that they are able to easily pass the famous Turing Test, which was proposed by the British mathematician Alan Turing in 1950 as a criterion for determining whether a machine is capable of exhibiting intelligent behavior indistinguishable from that of a human.

This test depends on the machine’s ability to conduct a conversation in natural language, so that a person cannot distinguish whether he is talking to another person or to a machine.

Among scholars artificial intelligence, There are few names with as much authority and perspective as Professor Hilbert, the world-renowned academic who has dedicated his career to unraveling complex digital social systems.

In the media, Hilbert was a leading voice in warning against Cambridge Analytica’s interference in Donald Trump’s campaign a year before the scandal broke. In addition, he was a pioneer in the academic field with the first study that determined the amount of information in the world. In the field of public policy, he designed the first digital action plan in cooperation with… Governments of Latin America and the Caribbean.

Recommended reading: Rocket Knight Returns: The story of the video game Rocket Knight Adventures takes place in modern times

In an interview with El Colombiano, Hilbert offers his vision on the present and future of artificial intelligence, highlighting the challenges and opportunities that this technology presents to society. Additionally, it highlights how generative AI, beyond being a tool of persuasion, is designed to improve our relationships and increase empathy and emotional intelligence.

Let’s start by talking about artificial intelligence and process automation using this technology. What are the benefits and risks?

“Around 2020, while we were mainly focusing on the pandemic, what is known as the knowledge processing parameter gap occurred. This means that the parameters in neural networks increased dramatically, leading to amazing progress. This was a surprise even to experts like me, because through Increasing data and computational power (i.e. parameters in neural networks, which represent the connections between nodes), enables us to create systems that are practically indistinguishable from humans.

How does this relate to previous advances in artificial intelligence?

“Alan Turing, considered the father of computer science, raised the fundamental question in the 1950s about when we can consider a computer to be truly intelligent. He concluded that if when dealing with something behind the scenes you cannot tell whether it is a human or a machine, then To accept that he has intelligence. “This way of assessing intelligence is similar to the way we evaluate human intelligence, in that we cannot directly observe the human brain to make this determination.”

Read also: AI tool to enhance restaurants in Medellin How does it work?

Do you think we have reached this point with current technologies?

“I always believed that we could eventually pass the Turing Test, which is what it’s called when a machine can pass a conversation like a human. I was surprised that this even happened in my lifetime, in 2023. We can now say that systems like ChatGPT are indistinguishable from humans in these interactions.” Even here at UCLA, where we assess intelligence through admissions tests like the SAT, the average IQ for humans is 65%, while ChatGPT’s is 90%.

What other areas are affected?

“For example, to be accepted into a doctoral program here in the US, ChatGPT outperforms humans on many academic tests. Even in fields like law and medicine, humans average 60% on these tests, while ChatGPT scores 85.” %.You can even get licensed as a primary care physician. “It’s interesting how the standards have changed in such a short time.”

Do you think technology will eventually impact our perception of ourselves as a species?

“We have to accept that we now live with machines that exceed the human average, according to all until recently conventional intelligence assessment criteria. This prompts deep reflection on our identity as Homo sapiens, a species known for its cognitive ability. “We are witnessing a profound change that deserves careful consideration. “

Let’s talk about AI beyond tools like chatbots, what are some lesser-known applications that are having a huge impact on various industries today?

“In the past 10 or 15 years, AI has mainly focused on machine learning using neural networks, inspired by the workings of the human brain. This has enabled the identification and generation of data, with increasing emphasis on the latter function, especially in modern business applications.

From image recognition to the recommendation algorithms used by giants like Google, Facebook, and Amazon, identification remains crucial. “Data generation is a more recent technological advance, which is transforming how AI capabilities are applied across sectors.”

What other types of artificial intelligence have an impact on industries other than the content industry or the artistic industry?

“Generative AI itself, like GPT, goes beyond chatbots and has different applications. In industries like digital business, and especially here in Silicon Valley where my students write code, tools like GitHub’s Copilot, derived from GPT, increase productivity by up to 55% for senior programmers who can earn half a million dollars a year.

This means that professionals who previously took five days to complete a job can now do so by Wednesday afternoon thanks to tools that are now practically within everyone’s reach.

You may be interested in: Without spending a penny: Learn how to expand the Wi-Fi signal in your home in 6 simple steps

What can you say about the ethical challenges facing the technology industry and the application of artificial intelligence in various fields, including personal data protection?

“This is an issue that worries me a lot. I remember that 20 years ago, in a philosophical conversation with a friend, we were imagining what it would be like if artificial intelligence actually existed. We were excited about the idea, but we were also aware of the potential risks.

At that time, we set three rules: don’t connect her to the open internet, prevent her from interacting with vulnerable people like children and those with mental health problems, and make sure she can’t program herself to achieve super-intelligence.

Unfortunately, we now see that we failed in all these precautions. The AI ​​is integrated into all open processes, interacts with children, and writes code more efficiently than many of my students. “We’re past the point where we can turn it off with a red button.”

So how do you see AI being integrated into many aspects of society and the economy today?

“This artificial intelligence, which surpasses elementary humanity in intelligence, is rapidly integrating into all areas: economics, education, and even people’s personal health. It knows more about our health than many doctors. This integration also includes areas such as politics and democracy, although we do not We fully understand its effects.

This complexity and lack of understanding is not new in the social sciences and adaptive systems such as economics or war. AI adds another level of complexity to these systems, and we must treat it with great care and respect, recognizing that we did not fully understand it from the beginning and understand it even less now.

What are the solutions to address these challenges?

“It is clear that we have been able to manage complex systems such as the stock market and democracy for centuries, despite not fully understanding them. With artificial intelligence, we must proceed with respect and caution, without falling into anxiety.”

The solution goes through two main paths. First, we must focus on mitigating the existential risks involved with this technology, and automating fundamental aspects of our species. This requires investing more resources into researching and understanding the impact of AI, as well as creating a supervisory algorithm to audit and control its use. Second, we must explore how to harness the benefits of artificial intelligence in a positive way, and promote its responsible integration into various social and economic processes.

What is your vision for the future of artificial intelligence and its impact on society?

“I believe the real value of models like GPT lies not only in their ability to generate interesting conversations, but in their ability to be professional agents for AI.

This is the new horizon: creating millions of agents with specific expertise, similar to the development of websites and mobile applications in the past. I envision a future where these agents are available for various functions such as marketing, education, health, software development, and more. These agents are programmed in a way that is accessible to everyone, not necessarily using code, which democratizes their use and makes them useful tools for improving quality of life and increasing productivity in multiple areas.