The stage of artificial intelligence

WhatsApp Group Join Now
Telegram Group Join Now
Instagram Group Join Now

One of the three films with the most Oscar nominations, it won statues for Best Picture, Best Director and Best Screenplay, among other awards. In a brilliant final scene, the screenplay hints at an impending shift in audience preferences: many of the dangers that threatened Margo at the beginning of the film now await Eva.

Digitalization, automation and robotization of processes in relation to artificial intelligence

We have been hearing about digitization, automation and robotization of processes for years. Recently, the emergence of artificial intelligence (AI) has pushed the former into the background, making its way into our daily interactions with the veiled promise of solving our needs quickly and easily. is: “Ask a chatbot”. Driven by such an effective narrative, more and more organizations with no prior experience in this area feel compelled to jump on the AI ​​bandwagon.

Such apparent simplicity is counterintuitive at first glance. As explained by Stuart Russell and Peter Norvig in their excellentArtificial Intelligence: A Modern Approach‘, artificial intelligence is the field of knowledge that seeks to design solutions, whether algorithms, robots or machines “…that can operate efficiently and safely in a wide range of new situations”. The promise of automatic adaptation to unpredictable situations is a distinct feature of AI compared to its predecessors such as automation or process robotization, which are also committed to eliminating human intervention, but through re-engineering to include new scenarios. Direct intervention is required.

In an increasingly sensitive environment, artificial intelligence plays the role of Eva in Mankiewicz’s film, while the automation and robotization of action, Margo Channing, her mentor: many of the dangers faced by the veteran diva will be inherited by her young. . and talented successors.

One of them is the risk of working on certain stages. Despite the advances made by artificial intelligence in recent years, neither machine learning nor, of course, its predecessors in viewer preferences guarantee a zero error rate. And this leads us first to consider their consequences depending on the scenario where our new star performs.

Imagine an AI competing in a chess tournament. Any potential mistake made by the AI, however embarrassing, does not jeopardize the integrity of the adversary or expose the team developing and training its models to potential legal action. leads to

Now let’s think about autonomous driving in a car: that’s a different story, isn’t it? In such a strange scenario, classical automation faced similar risks: this is why hybrid or supervised solutions are proliferating in sectors such as the automotive industry, where sensitivity is becoming increasingly important. Imagine, for example, parking in a modern car; If the sensors detect the proximity of an obstacle, the vehicle’s software will alert us to the potential danger. However, the final decision will be made by the driver.

We found the simplicity promised by AI to be counter-intuitive, as the training models on which it is based require a lot of effort on the part of properly qualified human personnel.

Today’s censored world helps amass a staggering amount of data, so two tasks become particularly important: data cleaning and data classification.

Data cleaning and classification

The need for these two activities can be better understood with the help of another analogy: the inclusion of video refereeing in decision-making during a soccer match. In particularly controversial plays, the referee may stop the match momentarily to review the available images taken from different angles with the help of the VAR, until he chooses the most appropriate one. It should be noted that a camera with the best possible orientation does not always help to remove suspense, for example, another player gets between him and the protagonist of the action. In the world of artificial intelligence, discarding shots would be part of data cleaning.

Even with video refereeing, the decisions a referee makes can be highly controversial, which is why they meet periodically to review pre-selected plays by a committee with the goal of establishing a common standard. have to do Some of these will also form part of the curriculum used to train new generations of referees. Returning to artificial intelligence, it is this role that classifies, because, based on a set of vague but documented conditions, a decision criterion is established with which to train the model for future decision-making. (and retraining) is given.

From the end user’s point of view, the promise of simplicity with AI is valid, because as a viewer they don’t understand this layer of complexity: cleaning and classifying data is a behind-the-scenes activity. But while the Margo Channings of our modern history require re-engineering, it is common for modern AI models to rely not only on initial training (the letter P in the Chat GPT acronym stands for ‘ is pre-training’) but rather successive re-training, through fine-tuning techniques that still require the involvement of human experts.

Other challenges of artificial intelligence

We conclude this quick review with another challenge that AI shares not only with its predecessors (automation and robotization of processes), but with any project. We are referring to the difficulty we humans have in defining and prioritizing our needs, as George H. Gallup anticipated nearly a hundred years ago when he founded market analysis. was Who has not suffered a major misunderstanding as a result of the process of grasping or interpreting the demands in his body?

But that’s another story that deserves to be told in more detail, and we’ll leave it at that for today, because as Margo Channing said:Fasten your seat belts, it’s going to be a rough night!

WhatsApp Group Join Now
Telegram Group Join Now
Instagram Group Join Now

Leave a Comment