Discover more from DSBoost
Is Data Science Still the Sexiest Job? - DSBoost #21
Welcome to the 21st issue of DSBoost, the weekly newsletter where you can discover interesting people in the ML/AI world, get the main takeaways of a relevant podcast, and stay up to date with the latest news in the field!
🎙️ Podcasts of the week
To move on AI projects is now. You don't want to just sit back and wait for things to develop. Start experimenting aggressively. Think big about what it's going to do for your organization and how it's going to transform it.
Is Data Science still sexy?
Data science is still sexy but it has become a lot more institutionalized. Companies are trying in all sorts of ways to understand their data and develop products and services based on it. But that's still a minority of companies. A lot of the decision-making is still not data-driven.
What are the biggest industries where data is widely adopted?
Financial services, both banking, insurance, and investments.
What are the obstacles for organizations in becoming more data-driven?
The main issue is culture:
It is coming from the top of the organization. If you don't have senior executives who are really committed to data science and analytics. It's gonna be much harder for the rest of the organization to adopt a data-driven culture.
There's a real imbalance between the attention that technology gets and the attention that the cultural side gets. Investing in cultural change and education is equally crucial for the successful adoption of data and analytics.
Thanks for reading DSBoost! Subscribe for free to receive new posts and support my work.
A 6-step framework for data-informed decisions:
Ask good questions
Acquire the right data
Do the analysis
Add personal experience
Make a decision
The 4 level of analytics:
Descriptive - What happened?
Diagnostic - Why the what happened?
Predictive - Take the what and the why and use them for predictions.
Prescriptive - Data tells you what you should do.
Most companies are good at Level 1, but they are stuck there. They try to do Level 2, but they are usually not good at it. And they invest all their money in Levels 3 & 4 because of the AI/ML hype. For companies, Level 2 would be the most important but it gets ignored! It gets ignored because they don’t know how to do it.
🧵 Featured threads
🤖 Top story of the week
Meta AI researchers have developed Voicebox, a generative AI model for speech that can generalize across tasks with state-of-the-art performance.
Unlike previous speech synthesizers, Voicebox does not require specific training for each task and can modify any part of a given sample. It uses the Flow Matching model, which allows it to learn from varied speech data without careful labeling.
Voicebox can perform tasks such as in-context text-to-speech synthesis, cross-lingual style transfer, speech denoising and editing, and diverse speech sampling.
The model outperforms existing models in terms of word error rate and audio similarity.
Although the model is not publicly available due to the risks of misuse, the research paper and audio samples are shared for further exploration and discussion in the AI community.