Artificial Intelligence

A few months ago, I received a Wunderman Thompson’s newsletter in which, as if it were a chapter in Black Mirror, it affirmed the existence of a mixed physical-virtual universe called Metaverse. With intrigue, I searched on the internet and found out that Facebook would become a Metaverse company rather than a social media one, so they were planning to hire 10,000 people to work in it. Another news said that the video game developer and publisher Epic Games, owner of the popular Fortnite, got a $ 1 billion investment to develop their Metaverse. These companies and many more have followed the same vision. So, what is the metaverse? Why are these tech giants so interested in it? What does AI have to do with all of this? Well, I will tell you how both humans and machines are shaping the Metaverse.

What is the Metaverse?

To give you an idea about Metaverse, I will give you an example with films such as Ready Player One or Avatar. In both, the characters live in an earthly world, like the one we live in. However, in both, the characters are aware that their reality has two worlds or realities which share the consequences of what happens in each other. For example, in Ready Player One, there is a virtual world called Oasis which you access through gadgets that let you live and interact with that world. In Avatar, there is the planet Pandora in which humans can transfer their consciousness through a capsule to a humanoid being living on the planet. These films show that what happens in Oasis or Pandora has a direct effect on reality, and it is because both Oasis and Pandora have been merged into a single entity with our world. Today there is no clear definition of the Metaverse, but for me, it is every experience that can be constantly shaped and replicated in both the virtual and the physical world, reprogramming beings and machines.

Ready Player One movie

These films can give a sense of some properties and characteristics that the Metaverse will address in the future and that has been mapped in a report by Wunderman Thompson, highlighting 9 key characteristics:

Mapping the Metaverse by Wunderman Thompson

All these 9 ingredients will allow us to go further in our world, from a passive participation of digital content to a totally active and inclusive one, where each action we make will affect our physical and virtual environment simultaneously.

A human – machine tendency

But why large companies have invested so much in the development of this Metaverse? It is because the increase in social networks, entertainment, and technological advances such Blockchain and AI that have being immerse in our everyday activities.

Number of people using social media platforms, 2004 to 2010 graph from the rise of social media report

The lockdown promoted and reassured the use of digital platforms as alternative means to entertain, socialize, shop, work and more. We have made these digital means part of us. This is reflected in data collected by Wunderman Thompson, where 61% of the surveyed say that their nowadays life is technology dependent. Jon Radoff, an American entrepreneur, explains the impact of the Metaverse in our lives through human-machine trends:

Virtual Mainstreaming: People tend to make the digital world as real as the physical world. If you’ve ever played Farmville on Facebook or were on Haboo making friends, you will sense the feeling of reality in these games. Or, to put other examples, Amazon adopted the way of shopping or Netflix watching movies from home. We unconsciously use the digital version of what we do physically in the real world, and we love it. As Beth Kindig writes in her article for Forbes: “If you experience moments where your virtual life online feels as real as your physical life, then you ‘ve dipped your toe into the idea of a Metaverse.”

Avatars at the disco in Haboo

Simulating Reality: As Jon Radoff mentions in his article, we will go beyond the Internet of Things (IoT) to an Internet of Everything. Technological advances let us think that, in the coming years, machines will be able to acquire and process huge amounts of data from our world in real-time, allowing the Persistent and Reactive properties mentioned in the Wunderman Thompson report. In addition, with the arrival of the Nvidia Omniverse, digital twins will allow us to simulate and explore better ways to optimize and operate real objects in much more controlled environments. This allows us to simulate how a factory operates like in BMW, how it feels to live in an apartment in downtown, or to generate military tactics.

Relationship between the real world and the Metaverse from Blockchain and AI Meet in the Metaverse

Machine Intelligence: AI as the mainstay of the Metaverse

To make real the Metaverse, we need machines to understand us and be reactive to our interaction, and AI has the potential to achieve this. In fact, AI will have a key role in being facilitator of interactions, participant within this universe and support for the human being, getting involved in processes of creativity, automation, security, and privacy, and becoming our best partner. Think of writers like J. K. Rowling, author of Harry Potter, who was able to construct a thrilling story in a magical world rich in detail. By reading or watching its movies, we may experience the Harry Potter universe to a certain extent. Instead, if we want to create our own universe and let people be part actively of it, AI will support us in creating compelling stories like we could do in AI Dungeon like tools, forming more realistic characters with Wizard Engine, or speeding up the construction of entire spaces with Promethean AI.

Moreover, it will not only be enough for us to create this new universe, but it should remain attractive and immersive enough so people want to participate and interact in it actively. AI allows us to collect key information that will help us understand our users better and continue to create memorable experiences for them. However, as it is sensitive information on the playground, the Metaverse will involve Blockchain. As suggested by the authors of the article Blockchain and AI Meet in the Metaverse, this technology mix will potentially ensure the transfer of data from each of the users to the Metaverse, to have traceability of who and why the user data has been used and maintain dynamic parameters that reduce vulnerability during encryption.

Illustration about the Promethean AI software

REFERENCES

Calandra, C. (2021, 1 julio). Entering the meta realm. Wunderman Thompson. Recuperado 2 de noviembre de 2021, de https://www.wundermanthompson.com/insight/entering-the-meta-realm

Newton, C. (2021, 22 julio). Mark Zuckerberg is betting Facebook’s future on the metaverse. The Verge. Recuperado 2 de noviembre de 2021, de https://www.theverge.com/22588022/mark-zuckerberg-facebook-ceo-metaverse-interview

BBC News. (2021, 18 octubre). Facebook to hire 10,000 in EU to work on metaverse. Recuperado 2 de noviembre de 2021, de https://www.bbc.com/news/world-europe-58949867

Epic Games. (2021, 13 abril). Announcing a $1 Billion Funding Round to Support Epic’s Long-Term Vision for the Metaverse. Recuperado 2 de noviembre de 2021, de https://www.epicgames.com/site/en-US/news/announcing-a-1-billion-funding-round-to-support-epics-long-term-vision-for-the-metaverse

Epic Games. (s. f.). Fortnite and Travis Scott Present: Astronomical. Recuperado 2 de noviembre de 2021, de https://www.epicgames.com/fortnite/en-US/news/astronomical

Wunderman Thompson Intelligence. (2021, 14 septiembre). New trend report: Into the Metaverse. Wunderman Thompson. Recuperado 2 de noviembre de 2021, de https://www.wundermanthompson.com/insight/new-trend-report-into-the-metaverse?j=61174&sfmc_sub=37405083&l=65_HTML&u=4069500&mid=110005021&jb=9008

Ortiz-Ospina, E. (2019, 18 septiembre). The rise of social media. Our World in Data. Recuperado 2 de noviembre de 2021, de https://ourworldindata.org/rise-of-social-media

Radoff, J. (2021, 1 noviembre). 9 Megatrends Shaping the Metaverse – Building the Metaverse. Medium. Recuperado 2 de noviembre de 2021, de https://medium.com/building-the-metaverse/9-megatrends-shaping-the-metaverse-93b91c159375

Kindig, B. (2021, 3 septiembre). The Key To Unlocking The Metaverse Is Nvidia’s Omniverse. Forbes. Recuperado 2 de noviembre de 2021, de https://www.forbes.com/sites/bethkindig/2021/09/02/the-key-to-unlocking-the-metaverse-is-nvidias-omniverse/?sh=62ca37795e17

NVIDIA. (2021, 1 noviembre). NVIDIA Omniverse^TM Platform. NVIDIA Developer. Recuperado 2 de noviembre de 2021, de https://developer.nvidia.com/nvidia-omniverse-platform

NVIDIA. (2021a, abril 13). NVIDIA Omniverse – Designing, Optimizing and Operating the Factory of the Future. YouTube. Recuperado 2 de noviembre de 2021, de https://www.youtube.com/watch?v=6-DaWgg4zF8

Safian-Demers, E. (2021, 25 marzo). Augmented living. Wunderman Thompson. Recuperado 2 de noviembre de 2021, de https://www.wundermanthompson.com/insight/augmented-living

Ankers, A. (2021, 19 octubre). The US Army Is Planning a Huge Experiment With Robot Tanks. IGN. Recuperado 2 de noviembre de 2021, de https://www.ign.com/articles/us-army-robot-tanks-experiment-planned?utm_source=twitter

Jeon, H., Youn, H., Ko, S., & Kim, T. (2021, 3 agosto). Blockchain and AI Meet in the Metaverse. IntechOpen. Recuperado 2 de noviembre de 2021, de https://www.intechopen.com/online-first/77823#B1

Fotos

Epic Records [Travis Scott]. (2020, 26 abril). Travis Scott and Fortnite Present: Astronomical (Full Event Video) [Vídeo]. YouTube. https://www.youtube.com/watch?v=wYeFAlVC8qU

Wunderman Thompson Intelligence. (2021a, septiembre 14). Mapping the Metaverse [Gráfico]. Into the Metaverse. https://www.wundermanthompson.com/insight/new-trend-report-into-the-metaverse?j=61174&sfmc_sub=37405083&l=65_HTML&u=4069500&mid=110005021&jb=9008

Our World in Data. (2019, 18 septiembre). Number of people using social media platforms, 2004 to 2018 [Gráfico]. The rise of social media. https://ourworldindata.org/rise-of-social-media

Haboo. (s. f.). [Avatars at the disco in Haboo]. Haboo. https://www.habbo.es/playing-habbo/what-is-habbo

Jeon, H., Youn, H., Ko, S., & Kim, T. (2021, 3 agosto). Relationship between the real world and the Metaverse. [Gráfico]. Blockchain and AI Meet in the Metaverse. https://www.intechopen.com/online-first/77823#B1

Promethean AI. (2021, 5 marzo). Promethean AI Keynote [Ilustración]. Promethean AI. https://www.youtube.com/watch?v=hA0MsGWvmzs

Ivan Caballero – AI Designer

VISIT OUR KNOWLEDGE CENTER

We believe in democratized knowledge

Understanding for everyone: Infographics, blogs, and articles

Check out our last articles:

Let’s tackle your business difficulties with technology

” There’s a big difference between impossible and hard to imagine. The first is about it; the second is about you “

Marvin Minsky, Professor pioneer in Artificial Intelligence

The ultimate goal of computational quantum chemistry is to tackle the quantum effects that determine the structure and properties of molecules. Reaching this goal is challenging since the energies associated with these effects are typically a tiny fraction of the total energy of the molecule. One of the applications of computational quantum chemistry is the study and design of drugs that block different stages of the virus’s life cycle. These types of drugs are called anti-retrovirals and they specifically bind with and block a virus protein or protease. With the protease blocked, the virus cannot make more copies of itself. It is important to perform chemical simulations to confirm that anti-retrovirals bind with the virus protein. However, such simulations are hard and sometimes ineffective on classical supercomputers.

Quantum chemistry for HIV

HIV is a virus that has presented a global challenge for public health. This virus has an impact on multiple societal dimensions including nutrition, access to health, education, and research funding. To compound the difficulties, the virus mutates rapidly with different strains having different geographic footprints.

Current supercomputers lack the ability to simulate HIV molecules and hence no treatment has been generated so far. However, quantum computers promise more accurate simulations allowing for a better drug-design workflow. For instance, the Variational Quantum Eigensolver (VQE) is an algorithm for finding the ground-state of a molecule and simulate other chemical phenomena.

How does the VQE algorithm work?

The inputs to the VQE are a molecular Hamiltonian and a parametrized circuit preparing the quantum state of the molecule. If you’re not a physicist, your most probable reaction will be: “what is that?!” Fortunately, you don’t need to know about quantum physics to understand VQEs. So, let me put it into other words. VQE is a hybrid quantum-classical algorithm, which means that the algorithm consists of two stages: a quantum and a classical stage. The output is an approximation of the combination of values that solve a given optimization problem.

During the quantum stage, a trial molecular state is created on the quantum computer. The trial state is specified by a collection of parameters which are provided and adjusted by the classical stage. After the trial state is created, its energy is calculated on the quantum computer. During the classical stage, a classical optimization algorithm looks at the previous energy levels and the new energy level and decides how to adjust the trial state parameters.
This process repeats until the energy essentially stops decreasing. The output of the whole algorithm is the final set of parameters that produces the winning simulation of our molecule and its chemical properties. Using this algorithm, scientists could find the anti-retrovirals that block the HIV virus among many other viruses that are computationally expensive to simulate.

But… what else?

Simulating molecules is only one of the multiple applications that the VQE algorithm offers. Its power is also extended to areas such as Machine Learning (ML) and Artificial Intelligence (AI). These fields rely on processing huge amounts of complex datasets. There is also a need to evolve algorithms to allow for better learning, reasoning, and understanding. While some ML and AI algorithms would take years in a classical supercomputer, a quantum computer would solve it in a matter of seconds using VQE’s. To wrap up, quantum computers are leading the way to the next generation of computers by increasing computational capability and power. With algorithms such as VQE’s, fields such as computational chemistry, AI and Machine Learning will enter a new era of power and speed.

REFERENCES

Pennylane.ai. 2021. A brief overview of VQE — PennyLane. [online] Available at: <https://pennylane.ai/qml/demos/tutorial_vqe.html> [Accessed 3 October 2021].
• 2021. [online] Available at: <https://towardsdatascience.com/the-variationalquantum-
eigensolver-explained-adcbc9659c3a.> [Accessed 3 October 2021].
• Qiskit.org. 2021. Simulating Molecules using VQE. [online] Available at: <https://
qiskit.org/textbook/ch-applications/vqe-molecules.html> [Accessed 3 October
2021].

Juan Francisco Rodríguez – Quantum Strategist

VISIT OUR KNOWLEDGE CENTER

We believe in democratized knowledge

Understanding for everyone: Infographics, blogs, and articles

Check out our last articles:

Let’s tackle your business difficulties with technology

” There’s a big difference between impossible and hard to imagine. The first is about it; the second is about you “

Marvin Minsky, Professor pioneer in Artificial Intelligence

Have you ever felt trapped in your tasks? You’re into a never-ending circle where as soon as you finish daily tasks you have to start all over with them. Those repetitive but also necessary activities have taken most daily time.

Robotic process automation (RPA) is a technology that develops robots. These robots are able to emulate some human interactions with different digital systems. And so can help with those routinary activities.

But how is it if robots are those big and heavy machines? well, here we talk about the informatics concept of software robots. They can do things like login into applications, extract information and report it. Additionally, it could be faster and has fewer errors completing the tasks.

How could it help?

When people think about automation usually appears an image of machines replacing humans. But nothing could be more unrealistic. The key concept of automation is to help. This rules-based technology will take and complete all those ‘boring’ and high-volume activities.

So, the less you have to stay on these tasks the more time you have. It gives people the time to use their intelligence and creativity to improve their skills and feel comfortable in their work.

Having robotic automation on the right processes could increase productivity and creativity and reduce mental stress. Now we will see a few tips to recognize those processes.

When to use it?

Something to keep in mind with RPA is not all processes can be automated. And some can be but are not useful. So, how do we know where it is useful? According to diverse authors, there are some important aspects to be considered:

It is a rule-based process (It follows a well-defined number of steps)
It repeats at regular intervals (once a month, every day) or has a defined trigger (every time an email arrives)
It has defined inputs and outputs (defined sources)
The RPA solutions will be a long-term solution (It will be executed several times)

And also, is important to identify if the original process needs a redesign. Because a human process with faults could also have an erroneous automation solution.

Some uses

In recent months, Equinox AI Lab has implemented this technology with clients such as Bayer and in marketing and human recourses processes. This has had an impact on reducing the time they spend looking for new opportunities or candidates. The automation gets and organizes the information from different sites and saves the filtered information. So, they have punctual information to check and save time.

Final thoughts and a step forward

As you may notice RPA is not necessary AI, it takes human process actions and completes them. But there is the concept of intelligent automation. This combines the best of both. RPA actions and AI processing information, thinking.

This gets a new vision of having characters or speech recognition and manages unstructured data. With this in mind, future use of RPA with AI could include read emails and give a specific answer related to the client petition. Or it is necessary to pass this information to a person who could give a better response.

In conclusion, RPA is a useful tool to reduce the time we spend in repetitive jobs with defined rules that can be modeled. Not all processes could be automated but the ones which can be, have a great impact on people’s lives and industry productivity.

REFERENCES

Blueprims. (s.f.). RPA AND INTELLIGENT AUTOMATION: A GLOSSARY. Obtenido de https://www.blueprism.com/resources/white-papers/what-is-rpa-what-is-intelligent-automation-heres-a-glossary-of-automation-terminology/

Casey, K. (30 de July de 2020). How to explain Robotic Process Automation (RPA) in plain English. Obtenido de https://enterprisersproject.com/article/2019/5/rpa-robotic-process-automation-how-explain

Marchuk, M. (18 de February de 2021). ¿Cuál es la diferencia entre la RPA, la automatización inteligente y la hiperautomatización? Obtenido de https://www.blueprism.com/es/resources/blog/whats-the-difference-between-rpa-intelligent-automation-and-hyperautomation/

UiPath. (s.f.). What is Robotic Process Automation – RPA Software. Obtenido de https://www.uipath.com/rpa/robotic-process-automation

Diana Perez – RPA Engineer

VISIT OUR KNOWLEDGE CENTER

We believe in democratized knowledge

Understanding for everyone: Infographics, blogs, and articles

Check out our last articles:

Let’s tackle your business difficulties with technology

” There’s a big difference between impossible and hard to imagine. The first is about it; the second is about you “

Marvin Minsky, Professor pioneer in Artificial Intelligence

What makes a number random?

Pick a random number between 1 and 10.

Was it 7?

Humans are really bad at picking random numbers. For instance, choosing 1 or 10 doesn’t seem so random because they are the largest and smallest numbers. A number picked near the middle intuitively feels more random than one at the higher or lower end. Even numbers seem less random than odd ones (though there is no reason for this to be true).

A true random number is equally likely to be any of the numbers and is completely independent of any previous number chosen. That means if you were to choose a large set of random numbers, each one would appear an equal number of times and it would be impossible to predict with absolute certainty the next few numbers. If you were to keep generating random numbers forever, you could produce any sequence of numbers (although this may take longer than the universe’s lifetime).

Why do we want random numbers?

For much of human history, random numbers were only used in games of chance. Dice go back 5,000 years (Piovano, 2011/2016). During WW2, random numbers became an important statistical tool for von Neuman when he was working on the Manhattan project (Metropolis, 1987) and for use by the Germans in sending encrypted messages.

From the Manhattan Project (Metropolis, 1987) came the Monte Carlo method. These are simulations that take a large number of samples from a model using random numbers to compute something that would be difficult to solve otherwise. They are a powerful and ubiquitous tool in physics, economics and data science. It became clear that random numbers were increasingly useful in areas of science, cryptography and statistics. More recently, with the abundance of private information sent over the internet, there is a considerable need to generate a large number of random numbers for various encryption standards (Zhou & Tang, 2011).

HOW DO CONVENTIONAL RNGs WORK

Computers generate random numbers in a deterministic way by taking a number that is close enough to being random (known as the seed) and then performing some iterative process to generate a sequence of numbers that appears random. For simple processes, the seed can be digits from the computers time in milliseconds.

Different algorithms generate the sequence in different ways. The middle square method used by von Neuman (Neuman, 1951) took a three-digit number, squared it, and took the middle three digits as the next number in the sequence. For instance, starting from 123, you compute 123²= 15129 and just take 512 as the next number and repeat as long as needed.

Another method, the linear congruential method (Thomson, 1958), generates the next term by calculating:

Where Xi+1 is the next term in the sequence, is the previous term, is the multiplier, is the increment and is the modulus. This sequence will repeat with a period less than m.

However, these methods are called Pseudorandom number generators (PRNGs) and do not actually produce random numbers. Anyone who knows the seed (initial random number) and the algorithm can generate the entire sequence with complete certainty. If left to run for long enough, both the middle square method and linear congruential method will repeat. This is fine for videogames where it is the feeling of randomness which matters, but not so great for encrypting communications (Li, 2013). With insider knowledge of the PRNG, an attacker could decrypt the communication.

QUANTUM RNG

Thankfully there are ways of generating truly random numbers based on physical processes. Atmospheric noise, the cosmic microwave background (the effect that caused static on old TVs) and radioactive decay are good examples. We can measure radio wave or microwave radiation, or the number of clicks from a Geiger counter.

Quantum computers can also be used for this purpose. They are effectively controlled physical experiments leveraging quantum mechanics to perform some computation. Since randomness is an inherent part of quantum mechanics, quantum computers, unlike classical ones used by von Neuman, can serve as a True Random Number Generator (TRNG) (Jacak, 2021). This is because a quantum system can exist in a superposition of possible states, and following a measurement takes on one of these states. Whilst we can know the probability of the system taking each of these states, we cannot know with absolute certainty which it will take (Nielsen & Chuang, 2000).

Below is an example of how n random numbers (from 0 to nmax) can be generated using IBM’s quantum computer. The code (can be found here) creates a quantum circuit with enough qubits to represent the power of 2 greater than nmax. The qubits are then put into a superposition and measured to obtain the random number. This process is repeated 1000 times and sampled n times to produce the numbers.

The TRNG implemented here is hardly the most practical implementation. It is rather slow, requires access to IBM’s cloud infrastructure, is vulnerable to interception, and runs on a device cooled to nearly absolute zero. A more useful implementation of this concept was done 20 years ago by the Swiss company ID Quantique, using a photonic chip. Newer models can be integrated in desktop PCs with PCIe connectivity or even USB (ID Quantique, n.d.).

REFERENCES

ID Quantique. (n.d.). Quantis QRNG Chip. Retrieved from https://www.idquantique.com/random-number-generation/products/quantis-qrng-chip/

Jacak, M. J. (2021). Quantum generators of random numbers. Sci Rep.

Li, A. (2013). Potential Weaknesses In Pseudorandom Number Generators.

Metropolis, N. (1987). The Beginning of the Monte Carlo Method. Los Alamos Science Special Issue, 15.

Neuman, J. v. (1951). Various Techniques Used in Connection With Random Digits. Res. Nat. Bur. Stand. Appl. Math.

Nielsen, M. A., & Chuang, I. L. (2000). Quantum Computation and Quantum Information. Cambridge University Press.

Piovano, I. (2011/2016). In Logic and Belief in Indian Philosophy. Warsaw Indological Studies.

Thomson, W. E. (1958). A Modified Congruence Method of Generating Pseudo-random Numbers. The Computer journal , 83.

Zhou, X., & Tang, X. (2011). Research and implementation of RSA algorithm for encryption and decryption. Research and implementation of RSA algorithm for encryption and decryption.

Thomas Clarke – Quantum Strategist

VISIT OUR KNOWLEDGE CENTER

We believe in democratized knowledge

Understanding for everyone: Infographics, blogs, and articles

Check out our last articles:

Let’s tackle your business difficulties with technology

” There’s a big difference between impossible and hard to imagine. The first is about it; the second is about you “

Marvin Minsky, Professor pioneer in Artificial Intelligence

Es un día lluvioso, te levantas muy temprano para ir a trabajar, y abres tu celular. Revisas tu correo, redes sociales y contestas algunos mensajes. Te arreglas para tu día, y mientras que ves las noticias te comes unos huevos con un café. Te alistas para salir, y pides un Uber para llegar al trabajo, mientras que escuchas Spotify o Apple Music y conoces tu nueva canción favorita. En solo de pronto dos o tres horas del día, has utilizado inteligencia artificial casi cada minuto de tu día. Desde la música que Spotify te recomienda mientras que te bañas o vas al trabajo, hasta los filtros de spam que utiliza Gmail o Outlook para evitar que veas esa propaganda en tu correo por veinteava vez.

La mayor virtud y utilidad de la inteligencia artificial no es lo que nos venden las películas, robots andantes que aprenden de nuestros movimientos para hacer una revolución. La inteligencia artificial, es lo que nos permite ser más eficientes cada instante, nos permite ser más asertivos al tomar decisiones, incluso más importante nos permite tener una mejor experiencia como humanos.

Cuando se desarrolla inteligencia artificial no solo se debe hacer en un esquema ético y legal, sino también se debe hacer con el propósito de complementar la inteligencia humana. Los humanos tenemos una capacidad mental inigualable, en últimas somos los creadores de la inteligencia artificial. Lo que nos permite esta nueva tecnología, es que esa inteligencia sea usada para innovar, crear, soñar, romper los límites de lo que pensábamos era imposible.

Si lo vemos en estadísticas “la inteligencia artificial tiene un margen de error del 3% mientras que los humanos lo tenemos del 5%” (Retina, 2019) en actividades automatizables como lo son complejos cálculos, análisis de datos, probabilidades, entre otros. Pero en contraste a esto, al comparar ambas inteligencias con actividades como argumentación, curiosidad, creatividad e inteligencia emocional, los humanos se llevan la delantera (Fresno, 2018). No se debe entender ambas inteligencias como contrincantes, sino como complementarias, como se han demostrado en diferentes áreas.

Un ejemplo se puede ver en el sector de la salud. Este trabaja todos los días para entender como el cuerpo humano y las enfermedades funcionan. Uno de estos es el cáncer de mamá, el cual es uno de los canceres más frecuentes con un porcentaje de 11,6% de todos los cánceres diagnosticados (Quironsalud). En mayo del 2019, el Computer Science and Artificial Intelligence Laboratory desarrollo un algoritmo que logra predecir la aparición de cáncer de mama hasta con 5 años de antelación (RGT Consultores Internacionales, 2020).

Incluso una alianza entre el MIT y ocho diferentes farmacéuticas que utilizan inteligencia artificial para sintetizar productos químicos y hacer estudios entre moléculas y funciones biológicas para crear medicinas en tiempo récord (RGT Consultores Internacionales, 2020).

Por otro lado, 15% de los niños en Colombia sufren de dificultades de aprendizaje, lo que les dificulta seguir su crecimiento profesional desde una edad temprana. Sin embargo, se ha creado una herramienta con IA que logra seleccionar recursos adaptados para este tipo de dificultades, así dando un enfoque especial para todos estos estudiantes (Inspiratics).

La revolución tecnológica ya ha empezado y va con mayor fuerza que nunca. Es una oportunidad para empresas, gobiernos y personas particulares de cambiar para mejor la forma en que viven y operan todos los días. Es la oportunidad de explorar todo lo que nos hace humanos, de una forma eficiente e innovativa. Confiar en esta tecnología, no es saltar a una piscina con los ojos cerrados sin saber lo que hay abajo, es saltar viendo el panorama completo. Mucho se ha desarrollado en Colombia y en el mundo sobre acuerdos legales para que el desarrollo de esta tecnología tenga un enfoque social y justo.

Adicionalmente, empresas líderes como Asesoftware, han tomado el liderazgo en este sector desarrollando cada día servicios y productos para que la mejor tecnología y el futuro llegue a tus manos.

¿Qué esperas para hacer parte del cambio?

REFERENCIAS

Retina, E. P. (2019, February 26). “La inteligencia artificial cuenta con un margen de error del 3%; los humanos, un 5%”. EL PAÍS. https://elpais.com/retina/2019/02/22/innovacion/1550841495_045791.html.

Fresno, B. G. del. (2018, May 21). Inteligencia artificial (AI): tres razones por las que los humanos son irreemplazables: BBVA. BBVA NOTICIAS. https://www.bbva.com/es/auge-maquinas-tres-razones-humanos-irreemplazables/.

RGT Consultores Internacionales. (2020, May 15). Inteligencia artificial en el sector salud. RGT Consultores Internacionales. https://rgtconsultores.mx/blog/inteligencia-artificial-en-el-sector-salud.

Inspiratics. (n.d.). 5 usos que ya tiene la Inteligencia Artificial en el aula. Inspiratics. https://inspiratics.org/es/recursos-educativos/5-usos-que-ya-tiene-la-inteligencia-artificial-en-el-aula/.

Quironsalud. (n.d.). Tipos de cáncer más frecuentes. Instituto Oncológico de Zaragoza. https://www.quironsalud.es/zaragoza-oncologia/es/cancer/tipos-cancer-frecuentes.

“Por medio del cual se establece la Inclusión Educativa de personas con Dislexia, Trastorno por Déficit de Atención con Hiperactividad –TDAH- y otras dificultades de Aprendizaje – DA”, PL 108-18, (2018)

Isabella Ochoa – Intern

VISIT OUR KNOWLEDGE CENTER

We believe in democratized knowledge

Understanding for everyone: Infographics, blogs, and articles

Check out our last articles:

Let’s tackle your business difficulties with technology

” There’s a big difference between impossible and hard to imagine. The first is about it; the second is about you “

Marvin Minsky, Professor pioneer in Artificial Intelligence

Todos tenemos una opinión frente a lo que es inteligencia artificial (en este blog, lo llamaremos AI). El cine y la televisión, como medios de comunicación masiva, nos ha empapado de historias y personajes relacionados a esta y han construido todo un imaginario frente al tema. Así, podemos tomar como ejemplo el universo cinematográfico de Marvel Studios en el cual nos presenta, por un lado, inteligencias artificiales como Visión, superhéroe cuya mente posee los restos de Jarvis, el asistente personal virtual de Tony Stark, y su contraparte infame Ultrón, robot cuyo objetivo es exterminar la raza humana.

En contraste, dentro de nuestra realidad, podemos pensar en sistemas de AI tales como lo son Siri o Alexa, presentes primordialmente en nuestros celulares para asistirnos, o las controversiales cámaras de reconocimiento facial implementadas en china, cuyo uso generó revueltas en Hong Kong por sentir que este tipo de tecnologías pueden llegar a vulnerar la privacidad de la población.

Con esto en mente, junto a nuestro equipo de AI llamado Equinox nos dimos la tarea de preguntarle a nuestros clientes, amigos y hasta familiares su opinión frente a esta tecnología. Si bien, muchas de las opiniones recalcaban ventajas como agilizar procesos, disminuir errores o procesar grandes cantidades de datos, de igual manera hacían énfasis en que esta tecnología competía, desplazaba y reemplazaba al ser humano, dejándonos a merced de esta sin ningún control. Debido a esto, me parece pertinente que nos preguntemos: ¿Está la Inteligencia artificial en beneficio a la humanidad? Antes de dar una respuesta a esta incógnita, debemos conocer un poco a nuestro sujeto de estudio.

Abrebocas a la AI

En las ciencias de la computación, la AI es la capacidad de una máquina de tener facultades cognitivas semejantes las de un ser vivo, especialmente a las del ser humano. Según IBM, empresa que lleva décadas trabajando en esta área, lo define como la simulación del pensamiento humano a través de modelos computarizados, los cuales le permiten entender, razonar, aprender e interactuar a través del uso de datos. Sin embargo, aunque las máquinas tengan estas facultades cognitivas, ellas realmente no pueden igualar en su totalidad a las de un ser humano. Asimismo, según Chris Noessel, el término AI podemos clasificarlo en 3 tipos: Débil, general, y fuerte.

La inteligencia artificial débil (weak AI o Narrow AI en inglés), es una AI que lleva a cabo tareas totalmente específicas como jugar ajedrez, dar recomendaciones como lo hace Netflix o poder hablar y dictarle comandos a nuestro celular a través de Siri. La inteligencia artificial general (AGI por sus siglas en inglés), es capaz de tener facultades cognitivas al mismo nivel que un ser humano, por lo que podría tener emociones y llevar relaciones.

Y, finalmente, una inteligencia artificial fuerte (ASI por sus siglas en inglés), es capaz de sobrepasar en todo sentido las capacidades de un ser humano. Por lo tanto, y a pesar de que en los últimos años hemos tenido grandes avances dentro de esta área, es claro que seguimos desarrollando AI débiles.

No obstante, ya hemos comprobado, para bien o para mal, el impacto que puede tener esta tecnología en nuestra sociedad. Por un lado, en el año 2014, la compañía estadounidense Amazon desplegó una AI encargada de llevar a cabo el proceso de reclutamiento y contratación de la empresa con el fin de automatizar y optimizar recursos dentro del proceso. Sin embargo, un año más tarde, el equipo se dio cuenta que esta AI tenía un sesgo a la hora de seleccionar los aspirantes, pues castigaba aquellas hojas de vida de mujeres y premiaba al de hombres. Otro ejemplo podemos encontrarlo en el documental Coded Bias, la cual expone que esta tecnología puede llegar a ser sexista, racista y opresora si no se entrena y desarrolla de manera correcta.

Por otro lado, tenemos los casos en los que los coches de la marca Tesla, los cuales cuentan con un sistema AI de piloto automático, se han visto envueltos en graves accidentes debido al mal uso de este sistema ya que se ha demostrado que estos sistemas no son capaces de responder eficientemente a las condiciones normales necesarias para una conducción real. Por ende, si estas AI débiles están teniendo un impacto adverso para el ser humano, ¿por qué continuar usando esta tecnología? ¿Qué mundo distópico nos espera cuando lleguen las AI fuertes?

La importancia del diseño en la AI

En el caso de Amazon, el equipo técnico y los diseñadores que construyeron esta AI debieron tener en cuenta qué partes de todo este proceso se podían automatizar, dónde intervenía el ser humano y no dejar que la máquina sea quien tome la decisión y defina quien debe ser contratado y quien no. Aun así, el equipo debió tener un entendimiento del contexto, ya que los datos con los que entrenaron a este sistema eran datos que demostraban el sesgo y la preferencia de contratación desigual en género que Amazon desde el 2014 quería empezar a combatir. Por otro lado, Josh Lovejoy en su artículo When are we going to start designing AI with purpose? describe los alcances actuales de la AI como “esenciales pero insuficientes por sí solas”, puesto que a las AI son una tecnología que tiene un aprendizaje similar a la de un niño. ¿Le daríamos total control a un niño para que maneje nuestros vehículos en las calles de nuestra ciudad? Esto nos lleva al caso de tesla, donde la falta de entendimiento de los usuarios hacia los alcances de esta tecnología, el poco análisis y prevención del comportamiento adverso que tendrían los clientes fue lo que ha ocasionado los graves accidentes que, en algunos casos, han causado muertes.

Repasando estos y otros casos de AI adversas se percibe levemente la falta de un componente que, de haberse tenido en cuenta, hubiesen tenido un mejor impacto: diseño centrado en el ser humano.

Si bien, los sistemas de AI han ganado popularidad abriendo las puertas a las distintas industrias de conocer a fondo a sus clientes, tener una mayor capacidad de análisis de estos para captar y mantenerlos, ¿Está la Inteligencia artificial en beneficio a la humanidad de los clientes, usuarios, seres humanos? Como lo describió mi compañero Manuel en su blog titulado Impacto del Design Thinking en las diferentes industrias, el uso de metodologías de diseño centrado en el humano permite abordar problemas complejos y que respondan a la pregunta: ¿cómo podemos hacer del mundo un lugar mejor? A través de esta podemos desarrollar inteligencia artificial en beneficio a la humanidad, con propósito humano y que trasciende de la perspectiva simple de negocio.

Usar esta metodología abre los ojos de quienes participan en el desarrollo de sistemas de AI donde, por un lado, nos hace conscientes para discernir cuándo interviene una AI, cuándo un ser humano, o cuándo deben colaborar ambos; y, por otro lado, permite planear y prevenir los posibles impactos éticos, morales, sociales, culturales y económicos que puede llegar a tener dentro y fuera de quienes usarán la AI.

La Inteligencia artificial en beneficio a la humanidad

Dentro de Equinox, nuestro grupo de AI, promovemos la colaboración equilibrada entre el humano y la máquina, la cual tiene como resultado no solo la automatización de tareas repetitivas, peligrosas o poco agradables para las personas, sino que pretende aumentar y potenciar nuestras capacidades humanas. Wildlife Insights es un proyecto que ejemplifica estos valores. Esta es una plataforma de Google en la que investigadores pueden monitorear, analizar y compartir datos en tiempo real sobre el estado de la vida silvestre a nivel global. Esta usa trampas y sensores para adquirir datos y usan AI para clasificar los distintos animales encontrados dentro miles de fotos, permitiendo a los científicos concentrar sus esfuerzos en planear y tomar acción para la conservación de los ecosistemas a nivel global. Otro ejemplo por destacar es Xray4All, una plataforma creada por investigadores de Stanford que proporciona análisis de radiografías al instante, indicando al usuario qué enfermedad ha detectado, en qué área concentra su predicción y con qué porcentaje de confiabilidad presenta su análisis.

Conclusión

El propósito de este blog era ir más allá de explicar o mostrar ejemplos de cómo con AI podemos captar y analizar a profundidad las dinámicas entre el cliente y nuestro negocio. Más bien, es una invitación a nuestros pares dentro de la industria y clientes que no debemos construir AI sin un sentido, o por moda. Quiero resaltar la importancia de construir AI con propósito y responsabilidad, manteniendo siempre el componente humano en mente. Finalmente, ¿Está la Inteligencia artificial en beneficio a la humanidad? Solo si desde ahora construimos nuestro futuro con responsabilidad.

REFERENCIAS

IBM Design for AI. (s. f.). Recuperado 18 de abril de 2021, de https://www.ibm.com/design/ai/fundamentals/
Education, I. C. (2021, 7 abril). Artificial Intelligence (AI). IBM. https://www.ibm.com/cloud/learn/what-is-artificial-intelligence#toc-what-is-ar-DhYPPT4m
(2018, octubre). Artificial Intelligence Driven Design (N.^o 1). Awwwards.books. https://www.awwwards.com/AI-driven-design
(s. f.). IBM100 – Deep Blue. Recuperado 18 de abril de 2021, de https://www.ibm.com/ibm/history/ibm100/us/en/icons/deepblue/
India, S. (2019, 5 noviembre). How Netflix’s Recommendation Engine Works? – Springboard India. Medium. https://medium.com/@springboard_ind/how-netflixs-recommendation-engine-works-bd1ee381bf81#:%7E:text=Netflix’s%20personalized%20recommendation%20algorithms%20produce,based%20on%20users%20viewing%20preferences.
Saini, K. (2021, 4 febrero). Natural Language Processing ft. Siri – MyTake. https://medium.com/mytake/natural-language-processing-ft-siri-2bc7b854a2a3#:%7E:text=Siri%20uses%20a%20variety%20of,(NLP)%20and%20speech%20recognition.
BBC News Mundo. (2018, 11 octubre). El algoritmo de Amazon al que no le gustan las mujeres. https://www.bbc.com/mundo/noticias-45823470
Bogen, M. (2019, 15 octubre). All the Ways Hiring Algorithms Can Introduce Bias. Harvard Business Review. https://hbr.org/2019/05/all-the-ways-hiring-algorithms-can-introduce-bias
Kantayya, S. (productor) y Kantayya, S. (director). (2020). Coded Bias [Documental]. EU.: 7th Empire Media.
Kolodny, L. (2021, 19 abril). «No one was driving» in Tesla crash that killed two men in Spring, Texas, report says. https://www.cnbc.com/2021/04/18/no-one-was-driving-in-tesla-crash-that-killed-two-men-in-spring-texas-report.html
Lovejoy, J. (2021, 22 enero). When are we going to start designing AI with purpose? https://uxdesign.cc/when-are-we-going-to-start-designing-ai-with-purpose-e196f986974b
Ortiz, M. (2021, 8 abril). Impacto del Design Thinking en las diferentes industrias – HOLISTIC – Design Lab. Holistic Design Lab. https://www.holisticdesignlab.com/design-thinking/impacto-del-design-thinking-en-las-diferentes-industrias/
Wildlife Insights. (s. f.). Wildlife Insights. Recuperado 18 de abril de 2021, de https://www.wildlifeinsights.org/home
Conservation International. (s. f.). Wildlife Insights. Recuperado 18 de abril de 2021, de https://www.conservation.org/projects/wildlife-insights#
(productor) y Wolff, T. & Yogeshwar, R. (directores). (2019). ¿De qué es capaz la inteligencia artificial? [Documental]. Alemania.: DW.

Ivan Caballero – AI Designer

VISIT OUR KNOWLEDGE CENTER

We believe in democratized knowledge

Understanding for everyone: Infographics, blogs, and articles

Check out our last articles:

Let’s tackle your business difficulties with technology

” There’s a big difference between impossible and hard to imagine. The first is about it; the second is about you “

Marvin Minsky, Professor pioneer in Artificial Intelligence