The A.I. revolution is here and I this is what I know about it

Everybody today is talking about the 2022 Football world cup... but something is changing history by the minute now... After many groundbreaking AI innovations have been released in the last year, one called ChatGPT was released 7 days ago and has now 1 million users. Human creative work will never be the same. Let me develop this in a 10 minute read...

I had been following Artificial Intelligence and thinking of sharing my views with my wide netowrk on this topic for a while. But the pace at which things are accelerating, inspired me to start writing just now.

My view below is not that of an expert in this subject, neither a scientist study. It does not represent the view of any of the companies I have worked at. It is just the humble opinion of someone that is fascinated about how technology applied to real life and business can change the world. So, if you are proficient in this subject, you don′t need to read further ;)

So what is Artificial Intelligence?

Artificial intelligence, is the simulation of human intelligence in machines to solve problems and perform tasks that would typically require human intelligence.

It encompasses a wide range of technologies and techniques including, but not limited to...

1. Natural language processing (NLP): this is how machines to understand, interpret, and generate human language. Applications: language translation, speech recognition, and text analysis.
2. Computer vision: this how machines understand and visual data, such as images and videos. Applications: facial recognition, object detection, and image classification.
3. Robotics: this is how machines move and act like living organisms. Applications: industrial automation, medical surgery, and space exploration.
4. Planning and decision making: this is how machines make decisions and take actions based on their goals. It is very much based on Machine Learning, which uses algorithms and data to improve performance. Applications: autonomous vehicles, supply chain management, and game playing.
5. Knowledge representation and reasoning: this is how machines represent and reason with knowledge and information. Applications: expert systems, question answering, and diagnostic decision making

For me the simplest way to see it is as corresponding to the respective human sensorial and mental functions.

Innovation accelerates at outstanding pace

I am always amazed how human kind is experiencing a incredible acceleration of its innovations. Since the beginning, it is all about how information flows amongst individuals, communities and the global population in time, which allows to further accelerate. The fire, was the driver of communities, who started the language. The wheel empowered travel, which empowered commerce and cultural exchange. Writing allowed to register and share knowledge. The print allowed to spread it at scale, etc. All this way to the age of internet, which started 40 years ago and was as rudimentary at the time as people connecting at super low speeds in stationary computers. Today, mobile data networks and smartphones allow for instantaneous flow of information worldwide.

Today, adoptions of innovations take place faster and faster. For reference, Gutemberg invented the printing press and died in poverty. It took 50 years for the press to be widespread in Europe and 450 years to go fully global. The internet took more than 10 years to become widespread in 1994-1995. Netflix took 3.5 years to reach 1 million subscribers. Facebook took 10 months, Instagram only 2.5 months... and ChatGPT (this mindblowing AI that I will describe below) took just 5 days to have 1 million users.

Artificial Intelligence is not new...

Artifical intelligence as a concept exists for over a century. In 1872, Samuel Butler's?novel "Erewhon" incorporates the notion of human like intelligence. In the cinematic universe it appeared in 1927 in the film "Metropolis". In scientific theory, Turing described in 1937 an abstract computing machine. The first real AI program was a system created in 1952 by Christopher Strachey and that ran in a Ferranti Mark 1 Computer.

Forbes has a pretty detailed timeline with key milestones of humankind did in A.I.

What is happening as we speak in Artificial Intelligence.

Until months ago, it was widely believed that machines could not enter the realm of creativity, let alone have feelings, empathy, etc. The first barrier of creativity has been breached. Apps released recently have proven that machines are actually capable of doing creative work in an extremely scalable and precise manner or replicate human behavior, voice and movements.

DIALOGUE AND CREATIVITY - ChatGPT

The most powerful app that displays this potential is ChatGPT. It is a conversational app that is able to maintain complex conversations, create content, interpret characters and moods, recognize and reproduce styles,.... all in a way you could not imagine. In my first 2 hours with the app, it provided me with the following within seconds and with super high accuracy:

• A Job description for a Web Analytics and CRO Specialist
• Structured arguments for a business decision
• Helped me with a legal doubt
• Recommendations on javascript coding
• Wrote a poem making fun of the Spanish soccer national team after disqualification from the world cup

The evolution of this app will have infinite capabilities that can enhance ANY intellectual / creative work that is non multimedia. It has applications in general on any business management task, arts, law, journalisms, medicine, psycology, software engineering, marketing, customer service, education...

It raises several questions about creativity. In the past it was consensus that creative jobs would be protected from robots and automation, as it would always be a unique human trait. But it seems creativity is just a very complex manifestation of intelligence, requiring good memory, capacity to relate concepts and do things differently. And a very advanced program has demonstrated it can do that.

When asked, it states that it does not have feelings, ethics, sense of humour or character. However, it can can dialogue portraying any given mood, character or even type of humour.

Apparently, in December 2022, Google reportedly issued a 'code red' amid the rising popularity of the ChatGPT AI

GRAPHIC CREATIVE APPS (DALL-E, MIDJOURNEY)

Another set of apps that are increasingly powerful can create realistic images and art from a description in natural language. Most desingers are using these applications to create infinite alternative creative designs.

VIDEO CREATIVE APPS (Runway)

These apps go one step further and allow you to generate and edit videos just with your text commands.

TEXT TO SPEECH AND VIRTUAL VOICE ASSISTANTS (murf, Amazon Alexa, Siri, etc)

These can go from are plain vanilla services to turn any text to natural voices... to voice assistants are able to reproduce someones voice in a very realstic manner. They have been conceived initially around home devices. This may be a limiting factor for concerns of privacy. Also, there are claims from experts that the usage is stagnant and displayed by discounts this past Black Friday season. Are people getting bored of talking to virtual assistants? Are virtual assistants solving a relevant problem? Is there one virtual assistant that will win over the rest?

REALISTIC HUMAN VIDEOS (Synthesia)

This is just a super powerful way to do videos based off text. Super helpful for demos, commercials or anything else.

PROGRAM CODING

Deepmind has created a system called Alpha code, that writes computer programs at a competitive level. AlphaCode achieved an estimated rank within the top 54% of participants in programming competitions by solving new problems that require a combination of critical thinking, logic, algorithms, coding, and natural language understanding.

What this means is that A.I.s can be creating other A.I.s. or developing themselves.

ROBOTICS:

There are several large robotics companies today. Just the top 10 are worth >$200B. Other may be smaller but are extremely innovative. Here are the top 35 most innovative robotics companies. My favorite one is Boston Dynamics, who has the most dynamic human-like robot.

MUSIC GENERATION

Here are 5 music generators. Not that it is relevant for all audience, but I think it is quite telling.

MANY MORE

Where can all this go?

I am not knowledgeable enough to comprehend where all this is headed. But my personal position is that the real impact will come as these technologies...

1. They become undistinguishable from human skills and enhance them at scale: All these apps have millions of users that keep training them on a daily basis. It wont take long for them to become quasi-perfect.
2. They get combined into powerful metasolutions: I have already seen good examples of people combining 2 A.I. apps and achieving extremely powerful things. Self-driving cars is one of the best examples of combining different technologies to emulate driving behavior. But imagine solutions that combine several technologies: image recognition, language, voice, reasoning, creativity, dynamics. They could eventually do anything a human can do. Since many of these solutions are Open source, combining them is quite feasible.
3. They start being applied by companies in commercial solutions: So far, the majority of corporates have struggled to find significant commercial value out of artificial intelligence. The largest tech companies, however are not only creating their own technology but will be the first ones investing in the cutting edge technologies.

What can companies can do about this?

Most of my friends working on companies outside of Tech giants say they are far from implemeting advanced A.I. Firstly, they claim they need to work on the basics and then sequentially climb the latter. The sequence could loook like this:

1. Digitalize your processes
2. Build a solid Data Deparment and Data Governance team
3. Start with business process optimization with machine-learning
4. Trial more advanced solutions

You could argue that you can bring external agencies to build sophisticated A.I. solutions, but it may be a challenge if they cannot rely on your data sets or your processes are not automated enough.

What can individuals do about this?

Do you have time for this?

Firstly, I guess you need to decide if this is a subject relevant enough for you to dedicate energy / interest on it. There are infinite hobbies and subjects to chose, and just in tech, there are multiple other trends, including Blockchain and Criptocurrency, Metaverse, Cloud Computing, etc.

Beyond my professional area of expertise and hobbies, there are 2 subjects I dedicate reasonable time to: Environmental Sustainability and Artificial Intelligence. I chose A.I. over the many other trendy technologies, based on a very simple mental framework. This is a very personal decision, as it relies on what you predict will happen in the future, what you believe is important, what you like and how you can act in your professional context.

In any case, for inspiration, Forbes just released the list of top tech trends to follow for 2023. Not surprisingly, "AI Everywhere" is the #1.

How do I keep updated?

There are many sources of information you can follow in digital media. I am quite a big user of Linkedin, so I follow a couple of linkeding groups on this topic.

Furthermore, your network of friends can be a good source. I do get updated on pretty useful stuff on AI, by my network of friends. The most valuable one is a whatsapp group I share with a cinema director, a creative director of Virtual Reality & inmersive experience, the VP of Product of a conversational technology (chatbot), a disc jockey (yes, a DJ), a Gen-Z geek lawyer and freelance writer, a politician and a guy that is off-network. It′s my favourite whatsapp group and keeps me close to (or away from) reality. I am also lucky to be friends with Elena González, one of the pioneers in language AI in Spain.

How do I get work exposure to artificial intelligence?

If you are self-employed, you are surely already using several productivity tools that are AI based. If you work in corporate, engaging with AI can be more difficult, but can also have more impact. If your company already has a Chief Data Officer, you should work with him on two things: firstly, building a solid data infrastructure of your business area. Secondly, identify critical business processes that can be substantially optimized with machine learning. You should offer your functional expertise for requirements and also for the model testing and validation.

I built my first robot when I was six years old, but actually it was made of carton boxes ;) I would love to see now how we conceived robots back in 1987, but I guess we expected them to have bow-ties.

At the professional level, I have been exposed to machine learning since ten years ago and mainly at Amazon. There, business employees either developed or used algorithms that could learn and improve performance in specific domains - the search algorithm, the demand forecasting algorithm.

In the creative side, I worked with Sriram Te, Hari Gurumurthy, Rahul Tamaskar, Shantanu Chakraborty and other colleagues in the India-based Data Science teams, on a machine learning model that would help identify which were the most relevant competitors in each country or product category, based on several attributes. to inform our pricing, selection and inventory strategy and allow us to be more competitive but at the same time more efficient. We used the model Random Forest and spent one year until we achieved a robust level. Later, we worked on predicting the sales potential of products not available yet on Amazon, based on publicly available information.?

Another good example is the Amazon search algorithm, on which I worked, in collaboration with the A9 team in Palo Alto in charge of search experience. I really praise the model of Amazon's search algorithm for products. While the basics of Amazon′s search engine are no secret in 2022 (multiple articles available for the public), it was a pioneer product search engine back in the time, that learned from the users. It could improve the search results for each keyword based on the behavior of consumers when they search that keyword. I remember when we were going to launch amazon.es in September 2011. The search results the day before going public were absolutely rubish. The night of the launch, the search expert in Europe, Gerd Flidner told me: "Don′t worry, a couple of days after we launch amazon.es, search results will improve drastically as consumers will be teaching the machine". This was more than 10 years ago and it blew my mind.

What should I know of machine learning before launching it in my business area?

From my experience, I learned that machine learning has some weakpoints. Its good to know about them for expectations management and to know how to focus:

1. Machines are as smart or as dumb as the quality of the information the consume and the rules or goals that you set. If the data is biased, incomplete, or otherwise flawed, then the model will produce unreliable results. For example, as a retailer, if you dont include in your demand forecasting algorithm, exceptional events that lead to abnormal demand (ex. a pricing error), next year, you may be buying too much inventory of that product. When working on the ML model for competitor classification we learned this first hand. Firstly, because of the need to have high quality of input information, which was hard to achieve (e.g. company revenue). Secondly, we had to do multiple iterations to find non-sense results, identify the root causes and modify the model.
2. Machines manage well events that repeat, but struggle with the unknown. Don′t expect machine learning models to help you in times of uncertainty, unless they are super sophisticated ;) For reference, I doubt that any machine was able to predict the social and economic effects of the COVID crisis in 2020, because it was an unprecedented one. Many models were able to predict the different infection waves, based on comparable pandemias and certain drivers. But, predicting how society, governments and company would react was much more of a harder job...
3. Machines may become a "black box" that no one understands. They are complex and hard to interpret. One of the few frustrating things I found in my last years at Amazon was that the employees running hte business were increasingly working with robots that optimized the strategic tasks, but users could not understand, explain or learn from the decisions of the machine.
4. Machines can be gamed by anyone that understands how they work. The best way to portray this is with videogames. Players that recognize patterns or flaws can game the system. I remember clearly the trick in the FIFA ′94 game to stand in front of the keeper. More practically, at Amazon, long after launching, I was responsible for Apparel in France, Italy and Spain, and was more experienced in search algorithms. I challenged the A9 team as I believed that the search engine could be manipulated by sellers and proved by placing a blue tie in the top of the search results of "dresses" in the Spain fashion store, just by playing with variables that any external 3P seller could employ. There I learned that ML tools need to also incorporate mechanisms to avoid others gaming the system.

Do I need to start studying coding tomorrow?

I don′t think it is absolutely critical to be a software programmer to be successful in a world that gets more and more automated. But having a level of affinity could be of good use.

... or should I focus on being more human?

What I do believe is that, in a world where the technical and creative skills may no longer be as differential, emotional skills will ever more important. Being able to partner, build trust, inspire, share, co-create. Already today, a rather common factor of failure of initiaves, projects and even companies, is non-technical: misalignment of objectives, lack of trust between teams, bureaucracy, lack of vision, egoes, etc. Successful leaders will handle this well.

In a world were robots become more relevant, I want to be more human.

At the beginning of my career I was inspired by professionals that were super smart, had unique skills or had profound knowledge, and I still admire them. Yet, in time, when I reflect back to my past experiences, the professionals who inspire me the most are ones that were able to extract the most out of you and make you believe in yourself, the one with infinite energy, the elegant one, the one that was always respectful, the humble one, the most passionate one, the one that never gave up, the one that was always willing to learn, the one that could navigate conflict, the one that was extremely persuasive, the one that was highly collaborative, the one that was generous with his peers, the most constructive one. Even collectives of people inspired me: probably the most effective team I have worked with, was a mix of individuals with highly complementary skills and interests, sharing exactly the same goal and having 100% trust.?

Should I get get concerned about all this?

I prefer to have an optimist stance in this matter. Humans have faced evolution during their entire existance. It has always been about adapting to the environment. A.I. may allow the society and the economy to be more efficient. It will certainly change the way we do things. But, from what we are experiencing today, robots may be fantastic on getting things done, but humans prefer interacting with people before any robot.

My vision is that humans will keep the tasks that are fulfilling and involve interaction with other humans, supported by robots.

One risk of automating intellectual work is that humans stop exercising mental functions, similarly to people that stop exercising physically in the era of automation in transport and physical work.

What should Governments do about this?

I still need to learn more from what governments are doing to address this topic. One thing I do believe is that a certain effort should be done on security, audit and ethics and that this requires international alignment. I also believe that the individual democracies will adapt by evolving regulations on labour, intellectual property and distribution of wealth.

Thanks for reading