“Success is not final, failure is not fatal: It is the courage to continue that counts,” said Winston Churchill. Over the last decade at Globant, we have been researching and developing technologies around artificial intelligence (AI). We have seen it progress with typical quasi-linear growth, sustained and slow, a common characteristic of the beginning of exponential curves. In recent months, this growth has entered the most accelerated part, surprising everyone and showing us a new way computers can interact with humans.
Today, new applications based on large language models (LLMs) such as Open AI’s ChatGPT, Google’s Bard, or GitHub Co-Pilot allow us to maintain an intelligent dialogue with them and see how they generate text and images in an almost human-like way. However, this brutal evolution comes hand in hand with new philosophical questions. For example, Yuval Harari, Tristan Harris, and Elon Musk, among other leaders, recently raised concerns through a public letter expressing the need to rethink the future of AI.
AI as a complement to humanity
One of the hottest discussions today is: Can AI replace jobs, and especially, as a novelty, knowledge-based jobs? A few days ago, OpenAI published a paper on this topic. What is happening now could resemble other technological changes that occurred in the past: the industrial revolution, where the machine replaced force, or, less impactful but more illustrative, what happened with the replacement of slide rules with calculators. In both cases, significant technological evolutions did not impact the importance of the work of engineers but improved their capabilities to carry out their tasks.
AI will accelerate the work of programmers, writers, and other creative jobs, but it still cannot replace them. It’s not for lack of capacity but because these systems lack information about the context we humans have. They cannot know what is happening in our community, work, or society, so it is impossible for them to generate the perfect text or code we were looking for at that precise moment.
There is another reason. These models have a great capacity to create dialogues and texts, but they do not have the ability to execute. Let me give you a simple example: Bard can write texts for an advertising campaign, but it is less likely that it can publish these texts in a medium and make a decision about the campaign. Not only that, but it would also have no responsibility for what it writes. These models can invent data or create incorrect information since they are neural networks that aim to make a semantically meaningful and coherent text but are not necessarily true. The lack of final responsibility for the information is undoubtedly a significant issue to solve, and it is a very big difference with humans. The fact that people trust these responses will be a major issue to be resolved since it is not easy to know the origin of the information that led to that response.
In short, we must imagine artificial intelligence as a personalized assistant that helps us improve our productivity. And together with this increase in productivity, there will be (as always in history) an increase in our ambition to build more things in less time. And now it will be possible.
Welcome to the real conversation
Following the massive landing of generative AI tools, we have seen a new way of interacting with machines, more conversational and less transactional. Today we can ask our favorite brands for a different, more human, more emotional, more aware interaction of each customer’s needs. And this applies to all companies, from telecommunications to banks, passing through the entertainment industry, travel, pharmaceuticals, and even industrial control companies. The dream of hyper-personalization is knocking at the door.
The LLMs are even challenging something that seemed impossible: how we search for information online. Successful models like Google’s are undergoing revision to find new ways of accessing information and creating texts without the need to navigate through thousands of links. Since ChatGPT was made available to the public, the battle for leadership among new ways of delivering search results has accelerated between major players like Microsoft and Google.
In this context, organizations are increasingly called upon to create AI applications and conversational interfaces combining large language models with their proprietary information. Channels like WhatsApp, for example, will become increasingly relevant.
The world beyond AI
Since the beginning of my career, I have witnessed the advancement of technology, from the arrival of the internet and social networks in the 90s and early 2000s, to web 3.0, the Metaverse, and artificial intelligence. All of these innovations have revolutionized the way we interact as humans. We can now communicate and connect with people worldwide more efficiently and easily. Web 3.0 has allowed us to carry out direct and reliable transactions, eliminating intermediaries, and the metaverse will give us a 3D digital space that could significantly impact how we work, learn, and entertain ourselves in the future. As I have mentioned, artificial intelligence has also progressed exponentially. And we will surely see even more significant disruptions, such as energy production through nuclear fusion or the massification of quantum computing, which will change security and processing speed paradigms. Technological development is truly in its exponential phase.
In the ever-expanding world of technology, humans will always remain at the center, with our abilities and flaws. One day, ChatGPT will be obsolete, replaced by a much larger and interconnected model. When that day comes, we will revisit what we think and what we know.
Click here to learn more about how Globant is thinking about the power of AI.