Steering the AI shift in business: a guide to seizing opportunities
Explore the four critical aspects of navigating the AI shift: Encouraging shadow AI, leveraging AI-enhanced tools, innovating with AI-driven products, and preparing for paradigm shifts.
Ciao,
any prediction about market trends, any competitive map or Gartner quadrant is about to be swept away by the overwhelming impact of generative AI. In this episode of Radical Curiosity I offer some thoughts on how to address the paradigm shift and perhaps seize the opportunity to create a new generation of products and services based entirely on artificial intelligence.
Table of content
Steering the AI shift in business: a guide to seizing opportunities
Black Box: a new podcast series about AI and us by The Guardian
The GPT-4 barrier has finally been smashed
Steering the AI shift in business: a guide to seizing opportunities
In the latest edition of Radical Curiosity, I wrote about the transformative role of artificial intelligence within the content localization and dubbing sector. Over the next five years, we can expect significant shifts:
There will be a transition towards automated translations, reducing the reliance on human translators. The role of linguists will evolve to focus more on overseeing, refining, and training these AI systems.
The integration of machine translation technologies will expedite businesses’ global expansion, allowing them to penetrate new markets by communicating in their customers’ languages.
Generative AI technologies will introduce novel service categories, including automatic dubbing using synthetic voices, enhancing accessibility to information and user experience.
The influence of artificial intelligence extends well beyond localization. Content creators are increasingly leveraging AI to enhance their productivity and creativity, utilizing it for tasks such as drafting social media posts, generating visuals, scripting videos, and even producing video content. This integration of AI not only streamlines the creative process but also unlocks novel possibilities for creators.
Simultaneously, digital platforms are rolling out innovative AI tools, offering creators fresh avenues to discover, interact, and disseminate content:
YouTube recently launched a suite of AI-powered features for its creators, including Dream Screen, a tool that integrates AI-generated imagery or video backdrops directly into YouTube Shorts.
TikTok introduced the Creative Assistant, an AI-powered aide designed to “support your video creation journey, helping you brainstorm ideas, understand best practices, uncover trends, and find inspiration.”
Meta unveiled new AI tools that can transform text prompts into a variety of unique, high-quality stickers within seconds. These stickers are readily shareable across WhatsApp, Messenger, Instagram, and Facebook Stories, further enhancing the social media experience.
The ubiquity of artificial intelligence is evident, as everyone, everywhere, is striving to comprehend its evolution, embrace AI-driven tools, or innovate with products founded on generative AI.
Recently, I had the privilege of speaking at a corporate convention, where the central discussion revolved around AI’s influence on digital marketing practices. I proposed an exploration through four distinct dimensions.
Promote Shadow AI
The rapid pace at which artificial intelligence is advancing is breathtaking, with major tech giants rolling out innovations on a near-weekly basis. I firmly believe in the critical importance of experimentation—whether subscribing to relevant services to catch a glimpse of the latest tech advancements, staying updated through newsletters, or tuning into podcasts. This kind of engagement shouldn’t just be a personal initiative; companies ought to encourage their teams to allocate 10 to 20 percent of their time (why not Friday afternoons?) to explore, experiment, and share discoveries with peers.
But encouraging this spirit of exploration also means meticulously crafting frameworks to navigate the intricacies of AI adoption. It’s about taking stock of which applications are being tinkered with and why, defining clear boundaries for the use of corporate data in external tools, and laying down simple, straightforward processes for employees to seek the clarifications and permissions they need.
Leverage AI-enhanced tools and services
Every business relies on software to operate efficiently. Given generative AI’s capacity to significantly streamline workflows, it’s crucial for leaders to ensure their tech partners are on the cutting edge. This involves rigorous comparison with the market and readiness to switch providers when necessary. Implementing software that adeptly incorporates generative AI can substantially boost productivity.
DevOps.com found that developers using Github Copilot see an average productivity increase of 5 percent—a figure that’s just the beginning (Best of 2023: Measuring GitHub Copilot’s Impact on Engineering Productivity).
In February, Klarna revealed its AI assistant effectively performs the tasks of 700 full-time employees (Klarna AI assistant handles two-thirds of customer service chats in its first month):
The AI assistant has had 2.3 million conversations, two-thirds of Klarna’s customer service chats.
It is on par with human agents in regard to customer satisfaction score.
It is more accurate in errand resolution, leading to a 25% drop in repeat inquiries.
Customers now resolve their errands in less than 2 mins compared to 11 mins previously.
It’s available in 23 markets, 24/7 and communicates in more than 35 languages.
It’s estimated to drive a USD 40 million profit improvement to Klarna in 2024.
Innovate with AI-Driven Products
It’s imperative for companies to envision a new wave of products that transcend merely incorporating artificial intelligence as an add-on feature. Consider the domain of translation, where machine translation isn’t a novel concept. Translated, for example, has integrated machine translation into Matecat, a tool for linguists, for years, continually assessing its efficacy. Matecat, alongside other tools in the industry, utilizes a sophisticated suggestion system partially powered by artificial intelligence.
Translated assesses technological progress using the Time to Edit (TTE) metric, which tracks how long it takes an expert translator to proofread another translator’s work. Typically, proofreading another expert’s work takes about one second per word (illustrated by the blue line in their graph), a rate that has remained steady over time.
In 2015, it required between 3 and 4 seconds per word for a linguist to review machine-generated translations. This review time has steadily decreased over the years, and it’s projected that machines will soon produce translations comparable to those of top human translators. Thus, the time needed to edit machine or human translations will converge, suggesting that machines are on track to match the translation quality of the best linguists in the field.
Until recently, the primary focus within the translation industry has been on utilizing artificial intelligence to enhance the efficiency of linguists’ work. Over the last ten years, this technology has seen constant improvements, leading to an increasing number of companies contemplating the shift from human to machine translation.
Having brought machine translation into the mainstream, the localization industry is now tasked with guiding its customers on the optimal use of this technology. This involves developing tools that assist customers in governing the shift and determining the appropriate instances and methods for employing artificial intelligence.
Prepare for the Paradigm Shift
In science, a paradigm shift, a term coined by Thomas Kuhn in his seminal work, The Structure of Scientific Revolutions, occurs when the accumulation of anomalies contradicts the prevailing understanding and cannot be explained within the framework of the current paradigm. Each paradigm has its anomalies, which are often overlooked or dismissed as acceptable deviations. However, when these anomalies become too numerous, they precipitate a crisis within the scientific community. During such crises, innovative ideas, including some that were previously rejected, are reconsidered. Eventually, a new paradigm emerges, obtains support, and ignites an intellectual clash between the adherents of the new and old paradigms.
What will the paradigm shift look like in the localization industry? It’s difficult to predict with certainty at this juncture, as we are only at the onset of a generalized crisis and reevaluation phase. There’s growing speculation around the potential for multilingual content generation. Should this direction prevail, traditional translation could become obsolete, shifting the emphasis towards systems and processes designed to verify the accuracy and pertinence of machine-generated content rather than focusing solely on translation quality.
My approach to preparing for the paradigm shift involves exploring possible futures, ranging from the most reassuring scenario (where everything remains largely unchanged) to the dystopian extreme (envisioning a future akin to humans in the Matrix, being used as batteries to power machines).
Black Box: a new podcast series about AI and us by The Guardian
I’m enjoying Black Box, a podcast from The Guardian, which, rather than zeroing in on the latest technology, offers listeners a wide-ranging and contemplative examination of the history and consequences of artificial intelligence:
At some point in the past few years, humanity collided with a new kind of intelligence. And things are getting strange. People are being accused of crimes by algorithms; falling in love with digital beings; pioneering new ways to fight old diseases; turning to machines for comfort in their worst moments, and using artificial intelligence to commit - and hide from - terrible crimes. The Guardian’s Michael Safi investigates the story of a technology so complex that its own creators have no idea what it is thinking, and captures a snapshot of the era when people first made contact with AI.
The GPT-4 barrier has finally been smashed
Four new AI models have emerged, challenging GPT-4’s top spot with impressive features and performance. These include Google’s Gemini 1.5, known for handling long texts and videos; Mistral Large with its strong performance; Claude 3 Opus, which excels in coding; and Inflection-2.5, which rivals GPT-4. Despite their success, these models are not open source and don’t share how they were trained, raising ethical concerns within the AI community.
Simon Willison discusses the emergence of ChatGPT competitors:
If you were forwarded this email or if you come from a social media, you can sign up to receive an article like this every Sunday.Thanks for taking the time to read this episode of my newsletter. I hope I’ve been helpful. If you think my sketchbook might interest someone else, I’d be glad if you shared it on social media and forwarded it to your friends and colleagues.
Nicola
Great effort put here, Nicola. Thank you for sharing your insights. Didn't read about Matcate till I read your article now.
Thank you for the great effort and valuable information.