# A year of artificial intelligence: the ChatGPT revolution and its global consequences

Author: Fernando Nieto Lobato
Original publication: 2023-12-06
Spanish original: https://estrategiabyaleph.substack.com/p/estrategia-10-chatgpt-y-la-era-de
English URL: https://elcontemplador.github.io/estrategia-english/essays/010/
Status: Published translation

This is a translation of the original Spanish essay published on 6 December 2023. Its claims, examples and forecasts retain that historical context.

English publication: 2026-09-29

*Historical context: this article was published on 6 December 2023, shortly after ChatGPT's first anniversary. Its assessments, figures and forecasts are preserved in that context.*

**Last week, ChatGPT celebrated its first birthday. In that year, artificial intelligence has moved rapidly and decisively from laboratories into the hands of billions of people around the world, probably driving the fastest global technological revolution in human history.** **Jim Fan**, one of the leading AI researchers at **Nvidia**, the company whose graphics cards are powering AI's expansion, summed it up eloquently in the following tweet. He was responding to an earlier post by OpenAI CEO Sam Altman recalling the night before ChatGPT's launch a year ago:

> **The past year in AI is like leaping from Stone Age to Space Age.** 🚀

[![Jim Fan compares the past year in AI to a leap from the Stone Age to the Space Age, quoting Sam Altman's recollection of the night before ChatGPT launched.](https://elcontemplador.github.io/estrategia-english/assets/images/62c197a6-b43b-4440-a310-4e9dd207a21e_901x394.png)](https://substackcdn.com/image/fetch/$s_!mgEY!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F62c197a6-b43b-4440-a310-4e9dd207a21e_901x394.png)

*The quotation above uses the English wording visible in the original screenshot. Sam Altman's embedded post reads:*

> a year ago tonight we were probably just sitting around the office putting the finishing touches on chatgpt before the next morning’s launch.
>
> what a year it’s been...

**At estrategIA, we want to use this first anniversary of ChatGPT**, the tool that decisively brought AI to a mass audience, **to look briefly at how the field has evolved over this period**. We will draw particularly on the “[Coatue Artificial Intelligence Full Report](https://www.coatue.com/blog/perspective/ai-the-coming-revolution-2023)”, a comprehensive analysis of AI's current state and future prospects, with numerous interesting charts. It was presented a couple of weeks ago by Coatue Management, a prominent global investment management firm known for its focus on technology.

As we mentioned, this is probably the fastest technological revolution in human history. These charts give an indication of its speed of adoption:

[![Coatue chart comparing US adoption of the Internet, smartphones and AI over time, with a dotted projection for AI extending towards 2025.](https://elcontemplador.github.io/estrategia-english/assets/images/9dd1ab0a-e8d8-484c-8718-eee6a8ffc488_1280x720.png)](https://substackcdn.com/image/fetch/$s_!ZCpU!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F9dd1ab0a-e8d8-484c-8718-eee6a8ffc488_1280x720.png)

*Chart text: “We're at Day 1 of AI… and riding on top of past waves.” The vertical axis measures US technology adoption as a percentage; the horizontal axis runs from 1990 to 2025. Blue represents the Internet, purple smartphones and green AI. The dotted extension of the AI curve is a projection, not observed adoption. Sources shown: World Bank and Statista, September 2023; Coatue analysis and opinion, November 2023.*

[![Coatue comparison of years to reach 50 per cent user penetration in the United States: PCs 20 years, Internet 12, mobile 6, and a tentative 3-year projection for generative AI.](https://elcontemplador.github.io/estrategia-english/assets/images/979057cb-99bc-4255-9d17-da45f1e2986b_1576x864.png)](https://substackcdn.com/image/fetch/$s_!vd9C!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F979057cb-99bc-4255-9d17-da45f1e2986b_1576x864.png)

*Chart text: “Adoption has been twice as fast with each platform shift.” Years to reach 50% user penetration in the US: PCs, 20; Internet, 12; mobile, 6; generative AI, 3, shown with a question mark. The generative AI figure is a projection. Source shown: Morgan Stanley Research, September 2023, reproduced by Coatue.*

We can also see very clearly ChatGPT's decisive importance in popularising AI. As a telling detail that reinforces its relevance, [only yesterday we learned that “ChatGPT” is Wikipedia's most searched-for term in 2023](https://www.telemundoareadelabahia.com/noticias/mundo/wikipedia-2023-articulos-mas-vistos/2358390/). Although growth has slowed considerably, its impact was enormous: global AI use went **from around 50 million users** — many of us were already using GPT-3 or the first publicly accessible image-generation AIs — **to 450 million in barely six months**, largely thanks to its launch. The following chart shows this clearly, along with some slowing after the initial surge.

[![Coatue chart of approximate monthly global AI application users, showing rapid growth after November 2022 and slower growth after May 2023.](https://elcontemplador.github.io/estrategia-english/assets/images/72a8a8f4-a014-40b9-8360-09cd5d146b33_1280x720.png)](https://substackcdn.com/image/fetch/$s_!SlC3!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F72a8a8f4-a014-40b9-8360-09cd5d146b33_1280x720.png)

*Chart text: “Has initial AI enthusiasm slowed down?” It plots approximate global AI application user counts by month from January 2022 to late 2023 and notes that growth has slowed since May. One annotation attributes over 60% of the user base to ChatGPT. Another offers possible explanations for slower usage: the student summer break and fading initial excitement as people learn which tasks AI handles well or poorly. These are the chart's hypotheses. Source shown: Coatue analysis and opinion, November 2023, covering a non-exhaustive set of applications.*

[The next chart, on monthly active users, was produced by the fund manager Global X](https://www.globalxetfs.com/chatgpts-one-year-anniversary-generative-ais-breakout-year/). It illustrates the year's evolution and events very well, both for ChatGPT and for the main moves by competitors seeking to keep up with the sector's leader.

[![Global X timeline plotting ChatGPT monthly active users from December 2022 to November 2023 alongside major generative AI product launches and investments.](https://elcontemplador.github.io/estrategia-english/assets/images/fd1bcced-5995-4c6c-b233-6289d302dc22_1600x1334.png)](https://substackcdn.com/image/fetch/$s_!Upcf!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffd1bcced-5995-4c6c-b233-6289d302dc22_1600x1334.png)

*Chart annotations, transcribed in readable form. These are the historical chart's claims, not newly verified figures. The vertical axis is monthly active users in millions; the chart labels the estimated total “200M+”. October and November 2023 are marked as forecasts.*

| Date shown | Annotation |
|---|---|
| December 2022 | ChatGPT exceeds 1 million users in five days, breaking Instagram's previous record of 2.5 months. |
| January 2023 | ChatGPT passes MBA and bar examinations. Microsoft invests US$10 billion in OpenAI. |
| February 2023 | Google takes a 10% stake in Anthropic; ChatGPT Plus is released; Meta unveils LLaMA; Microsoft integrates GPT into Bing Search. |
| March 2023 | GPT-4 is released. Google launches Bard and Salesforce unveils EinsteinGPT. |
| April 2023 | Amazon unveils Bedrock. |
| May 2023 | Meta unveils “AI Sandbox”, enabling personalised AI-based advertising. The ChatGPT app launches on iOS. |
| June 2023 | McKinsey estimates generative AI's economic potential at US$4.4 trillion per year. Nvidia joins the trillion-dollar valuation club. |
| July 2023 | OpenAI unveils Code Interpreter; Anthropic releases Claude 2; Bard becomes multilingual; Meta unveils LLaMA 2. |
| August 2023 | Meta introduces Code LLaMA, a state-of-the-art language model for coding. |
| September 2023 | Amazon invests US$4 billion in Anthropic and incorporates its technology into various products. |
| October 2023 | OpenAI exceeds an annualised revenue rate of US$1.3 billion; Microsoft GitHub Copilot exceeds US$100 million in annual recurring revenue. |
| November 2023 | OpenAI introduces GPT-4 Turbo and custom GPTs. Elon Musk's new AI venture, xAI, introduces its chatbot Grok. |

*Sources listed in the chart: Global X ETFs, drawing on TechCrunch (7 November 2023), Similarweb (September 2023) and McKinsey & Company (25 August 2023).*

The year 2023 has seen important developments in generative AI, a constantly evolving sector. Although OpenAI, with ChatGPT, retains a dominant market position, its leadership was temporarily affected by its CEO's recent departure and subsequent return, [officially confirmed a few days ago](https://openai.com/blog/sam-altman-returns-as-ceo-openai-has-a-new-initial-board). We analysed those events in detail in this newsletter two weeks ago. OpenAI is not the only major player, however. Technology giants including Google, Microsoft, Anthropic and Meta also play fundamental roles.

Google introduced Bard in March 2023, positioning itself as OpenAI's main rival. Originally designed as a research assistant, Bard has grown stronger following ChatGPT's integration with the Internet, adding Google's own resources, including an innovative YouTube extension in September 2023. Anthropic, meanwhile, launched Claude 2, also in March 2023, distinguishing itself through its focus on user safety and less harmful responses. Until recent weeks, Claude 2 had consistently ranked second in most model benchmarks.

Other notable competitors include Microsoft's Bing Chat, based on GPT-4, and Meta's Llama 2, an open-source large language model (LLM) with between 7 billion and 70 billion parameters. Llama 2 has inspired the development of numerous open-source models, driving an innovative field in which improvements appear weekly. Some of these models have surpassed GPT-3.5 and are moving ever closer to the market leader, GPT-4.

Recently, Elon Musk announced Grok, a new generative AI seeking to compete in this market. These innovations — even before considering advances in creating images, videos and music — demonstrate growing diversification and specialisation in generative AI, marking a period of unprecedented expansion.

Meanwhile, OpenAI also made major advances during 2023, particularly with the launch of GPT-4, a model far superior to its predecessor, especially in reasoning, and still the model to beat. It integrated DALL·E 3 for image generation and enabled the creation of AI proto-agents: the GPTs we discussed last week.

## Powerful changes at work, including in consulting

The figures above make clear the speed of the transformation AI is bringing. Beyond the new models, major technical advances and staggering investment figures, **analyses and data are beginning to emerge on AI's integration into work and its capacity to transform fields such as those covered by this newsletter**. For example, the Coatue report includes this interesting chart from [a study by Harvard Business School and the consultancy BCG](https://www.hbs.edu/faculty/Pages/item.aspx?num=64700). It makes clear the leap in consulting capabilities that AI brings even in its current, still very primitive state: improving the number of tasks completed, the speed of completion and, probably most importantly, the quality of the final result.

> “For each one of a set of 18 realistic consulting tasks within the frontier of AI capabilities, consultants using AI were significantly more productive (they completed 12.2% more tasks on average, and completed tasks 25.1% more quickly), and produced significantly higher quality results (more than 40% higher quality compared to a control group).”

*This quotation already appears in English in the Spanish article and is retained as supplied.*

[![Coatue chart comparing BCG consultants using GPT-4 with a control group: task completion rate 1.12 times, completion speed 1.25 times and quality 1.40 times the control baseline.](https://elcontemplador.github.io/estrategia-english/assets/images/99d7031d-ab51-43a2-9f7e-8247093b158e_1280x667.png)](https://substackcdn.com/image/fetch/$s_!0n24!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F99d7031d-ab51-43a2-9f7e-8247093b158e_1280x667.png)

*Chart values: the control group is set to 1 for every measure. With GPT-4, task completion rate is 1.12×, completion speed 1.25× and completion quality 1.40×. The vertical axis contains a break. The chart identifies Harvard Business School and BCG.*

Another powerful finding, closely connected to the educational purpose of Institución Educativa ALEPH and this newsletter itself, emerged from [a recent Salesforce study](https://www.salesforce.com/news/stories/ai-at-work-research/), whose title the article renders in Spanish as *Las Promesas y Dificultades de la AI Generativa en el trabajo* — “The promises and difficulties of generative AI at work”. It surveyed more than 14,000 workers in 14 countries, including Spain, and concluded that **more than a quarter of workers — specifically, 28% — already use artificial intelligence at work. In many cases, more than half, they have done so using tools not approved by their company and, above all, without specific training.** Given AI's enormous productivity benefits in many areas of knowledge work, employees will clearly turn to it quickly. It also seems highly advisable that they do so in an organised way and, above all, with the training needed to make the most of it and avoid many of the risks associated with such a new technology.

Those of us involved in education should also bear in mind that **future generations will use AI entirely naturally at work**. [A study published on the Nature website last month, concerning university students in Germany](https://www.nature.com/articles/s41599-023-02304-7), found that **more than two-thirds were already using artificial intelligence in their studies**.

ChatGPT's launch has dramatically accelerated the AI sector, giving it a pace beyond that of other enormous technological revolutions. Yet it is particularly important to remember that, although the present is already profoundly transformative, **this technology is still in the early stages of deployment** — even if its scientific origins go back more than 50 years — and **has enormous room for improvement**, at speeds that humans will struggle to process. We will probably devote another issue to AI's immediate future, again drawing on interesting charts from the Coatue report. But I would like to close with this one, which shows precisely the vast capacity for improvement and learning in the AI systems we are only just beginning to interact with.

[![Coatue chart of benchmark performance approaching or exceeding a human-parity line across handwriting, speech and image recognition, reading and language comprehension, common sense, academic subjects and code generation.](https://elcontemplador.github.io/estrategia-english/assets/images/ad239a50-3f5a-42f1-8182-2725e10c8338_1280x720.png)](https://substackcdn.com/image/fetch/$s_!Oe2A!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fad239a50-3f5a-42f1-8182-2725e10c8338_1280x720.png)

*Chart text: “We are optimistic: AI is getting better, faster.” It presents benchmark trajectories from 1998 towards 2024 against a horizontal “Human parity” line. The labelled tasks are handwriting recognition, speech recognition, image recognition, reading comprehension, language understanding, common sense completion, high school and college subjects, and code generation. The latter three remain below the parity line in the displayed endpoints. The chart's stated message is that the time taken to reach human-level benchmark accuracy is shortening. Source shown: directional data from Contextual.ai, Time and PapersWithCode, November 2023. No additional numerical scores have been inferred from the curves.*

Fernando Nieto Lobato

*Director of Digital Innovation, Institución Educativa ALEPH*