The Inktelligence - February 10, 2025

AI Isn’t the Problem—Leadership Is

AI is here, it’s powerful, and it’s reshaping work. But most companies are lagging. 92% of organizations are increasing AI investments, yet only 1% have actually integrated it at scale. The issue isn’t the technology or employees—it’s leadership.

• Employees are already 3× more engaged with AI than execs realize.

• 70% expect AI to automate 30% of their work within two years.

• 47% of leaders admit their AI rollouts are too slow.

Want to Win With AI? Stop Playing It Safe

The real challenge isn’t technology—it’s lack of urgency and execution. AI pilots are stalling because leaders aren’t thinking big enough or moving fast enough. Even in industries like healthcare and tech, where AI investment is high, progress is slow.

• Only 19% of execs are seeing real revenue impact from AI today, but 87% expect it to drive growth within three years.

• AI needs to move beyond experiments and be embedded into core workflows.

• Training employees to use AI effectively is just as critical as the technology itself.

Source: Superagency in the workplace, McKinsey report

Sam Altman’s Three Observations

In his latest blog post, Sam Altman talks about how AI is improving much faster than most people realize and what that means for the future. Here’s what he’s saying:

1. AI Gets Smarter as We Put More Into It

The more computing power, data, and investment we put into AI, the smarter it gets. The gain in intelligence is roughly the log of the resources put into training and running the AI.

2. AI is Getting Cheaper Very Quickly

The cost of using AI is dropping dramatically—getting 10× cheaper every year. This means that AI, which used to be expensive and only available to big companies, is becoming affordable for everyone, just like smartphones and the internet did.

3. AI’s Impact Grows Faster Than We Expect

As AI gets smarter and cheaper, its impact doesn’t just grow a little—it explodes. A small improvement in AI can lead to huge benefits for businesses, industries, and even entire economies. Companies that use AI well will gain a big advantage, while those that wait may struggle to keep up.

What This Means for the Future

• AI assistants will soon be able to handle real work, not just simple tasks.

• The world won’t change overnight, but in a few years, AI will transform industries just like the internet did.

• The most important skill? Being adaptable. The people and businesses that learn how to use AI effectively will have the biggest opportunities.

Altman’s main message: AI is moving faster than most people think, and those who embrace it early will benefit the most.

Read his entire blog post.

How AI impacts the workforce

In light of Cognizant's recent report, leaders must proactively assess how generative AI will reshape their organizations. The main takeaway for leaders is to be thinking about how their workforce will be shaped in the future.

• Fundamentally Unchanged: Roles where core tasks remain largely unaffected by generative AI, maintaining their current structure and function.

• Augmented: Positions where generative AI enhances certain tasks, leading to moderate changes in job functions and increased efficiency.

• Transformed: Jobs experiencing significant alterations due to generative AI, necessitating substantial reskilling and adaptation to new workflows.

• Fully Automated: Roles that could be entirely automated by generative AI, requiring redeployment or upskilling of affected employees to new positions.

Developing comprehensive talent management strategies is essential, focusing on reskilling and upskilling employees to adapt to new roles created by AI advancements.

L'Intelligence Artificielle Française: Mistral AI's Latest Innovations

Mistral AI has released a new AI model, Mistral Small 3 (available under Apache 2.0 license), a 24B-parameter open-weight language model designed for low-latency inference. It runs under Apache 2.0, achieves 81% on MMLU, and processes 150 tokens per second.

• It competes with models three times larger, including Llama 3.3 70B, Qwen 2.5 32B, and GPT-4o-mini, while maintaining a 32K context window for extended input handling.

• Key highlights include: 24B parameters with a 32K context window; three times faster than Llama 3.3 70B on the same hardware; optimized for latency, achieving 150 tokens per second; no synthetic data or RLHF, making it a strong base model for further fine-tuning; runs efficiently on local hardware, including an RTX 4090 or MacBook with 32GB RAM.

Mistral AI has also released Pixtral Large, a 124-billion parameter multimodal model that processes text and visual data. It builds on Mistral Large 2 and incorporates a 123-billion parameter text decoder and a 1-billion parameter vision encoder.

• The model supports a 128K context window, capable of handling 30 high-resolution images or a 300-page book.

• It outperforms GPT-4o, Gemini 1.5 Pro, and Claude-3.5 Sonnet on benchmarks like MathVista (69.4% accuracy), ChartQA, and DocVQA.

Coca-Cola is at the forefront of AI in marketing

Coca-Cola has fully embraced AI to revolutionize its marketing strategies, using generative AI, real-time data analysis, and interactive tools to engage consumers like never before. By integrating AI across creative, interactive, and data-driven initiatives, Coca-Cola is setting a new standard for AI-powered brand engagement. Though facing some backlash for their holiday ads generated with the assistance of AI, I think Coca-Cola will be way ahead of the competition as they are willing to experiment and learn from what works.

How Coca-Cola Uses AI in Marketing:

• AI-Generated Ad Content – Coca-Cola utilizes generative AI to create dynamic, personalized advertisements, blending real-time data with storytelling to produce hyper-relevant visuals and messaging.

• Personalized Consumer Experiences – AI-driven data analytics platforms analyze consumer behavior and preferences, allowing Coca-Cola to deliver tailored experiences across digital and physical touchpoints.

• Interactive Campaign Elements – AI-powered tools enable consumers to co-create content, such as custom Coke bottle designs and augmented reality (AR) experiences, fostering deeper engagement.

You don’t need to have the budget of Coca-Cola to make AI work for you. Start small, move fast and break things. It’s all part of the learning curve.

Apple just extended Pixar’s robotic lamp

Finally, something fun from the Apple research team. They made this articulated lamp IRL that reminds me of the one you see on the Pixar movie trailers. There is a video included with the paper below.

That’s it for this time. If you enjoy this newsletter, please share it with a friend.

Till next time,

Hock