Trends #008 - Week 8 - 2024
Gemini embarasses, NVIDIA can't stop, Nikkei 225 back on top after 34 years & Airflow
Hello Everyone,
Welcome back to TRENDS where we talk about the latest happenings of AI, the Tech Economy & latest Analytics tools!
Let’s see what happened in the tech world last week…
AI news 🤖
1) Google explains Gemini’s ‘embarrassing’ AI pictures
Probably the biggest AI news of past week!
Google's Gemini AI tool generated "embarrassing and wrong" images, such as racially diverse depictions of Nazis and US Founding Fathers.
Google attributes these inaccuracies to tuning issues in the model, causing it to "overcompensate" and become "over-conservative" in its responses.
Google's senior vice president, Prabhakar Raghavan, acknowledges the shortcomings in a blog post and apologizes for the feature not working as intended.
Raghavan emphasizes that errors are a known challenge with large language models, and Google is committed to addressing and improving them over time.
Gemini AI has temporarily disabled image generation capabilities in response to these issues.
Have you faced similar problems with other image generators?
2) Reddit is reportedly selling data for AI training
Reddit has reportedly negotiated a content licensing deal allowing its data to be utilized for training AI models, as per a Bloomberg report.
The deal, valued at $60 million, involves an undisclosed major AI company, coming just before Reddit's potential $5 billion initial public offering (IPO) in March.
The move aims to exhibit potential revenue streams in the rapidly growing AI industry to potential investors.
Reddit's vast collection of user-generated content, including posts from popular subreddits, comments from various users, and discussions on diverse topics, could be utilized for training and enhancing existing large language models (LLMs) or developing new generative AI systems.
Recent controversial decisions by Reddit, including removing private chat logs, implementing automatic moderation features, and removing the option to disable personalized advertising, have fueled discontent among users.
The reported deal to sell Reddit's data for AI training could further provoke backlash, contributing to ongoing debates over the ethical use of public data for training AI systems across various platforms and industries.
3) Mutale Nkonde’s nonprofit is working to make AI less biased
Mutale Nkonde is the founding CEO of AI for the People (AFP), a nonprofit aiming to increase Black voices in tech and combat algorithmic bias.
Nkonde's interest in AI was sparked by incidents of algorithmic bias, such as Google's labeling of Black people as gorillas in 2015.
She played a significant role in introducing the Algorithmic Accountability Act to the U.S. House of Representatives in 2019, establishing AFP as a key advocate for reducing algorithmic bias. Nkonde emphasizes the importance of women finding a niche in AI and becoming experts in that area to build credibility.
She highlights the need for more diversity in AI development and the creation of inclusive datasets to ensure AI systems work effectively for all demographics.
Nkonde suggests investors consider demographic trends and algorithmic bias during due diligence to push for responsible AI investment.
She advocates for reskilling programs to include marginalized communities and utilize their insights to ensure AI benefits everyone.
4) Nokia Partners With Nvidia on AI in Mobile Networks
Nokia and Nvidia have partnered to develop artificial intelligence solutions for mobile networks.
Nokia will utilize Nvidia processors along with its software to advance cloud network technology and explore AI's transformative potential in the industry.
The collaboration aims to deliver added value to end-users by introducing innovative AI services in the telecommunications infrastructure and services provided by mobile operators.
TECH ECONOMY news 🌍
1) Nvidia Sales Reach New Heights as Company Forecasts Bigger AI Boom
Nvidia released earnings last week where the company forecasted sales of around $24 billion for Q1 2024, indicating strong demand for AI.
Chief Executive Jensen Huang stated that AI has reached a tipping point, with demand for computing power underlying AI remaining high across industries, companies, and nations.
Nvidia's fourth-quarter results showed a significant increase in sales, with revenue more than tripling from a year earlier and earnings surging more than eightfold.
The company's success reflects the ongoing AI boom, with big tech companies like Microsoft, Google, and Apple investing heavily in AI technologies that rely on Nvidia's hardware.
2) Booking.com faces fines of almost half a billion euros in Spain
Booking.com faces a potential fine of €486 million for alleged anti-competitive practices in Spain, which would be the largest penalty ever imposed by Spain's competition watchdog, the CNMC.
The investigation, initiated in 2022, focuses on whether Booking Holdings' Dutch subsidiary has abused its dominant position in the online hotel booking market.
Complaints from the Spanish Association of Hotel Directors and the Hotel Business Association of Madrid prompted the CNMC probe, accusing Booking.com of practices that could be considered an abuse of its dominant position, including imposing restrictions on hotels' marketing activities and charging high commissions.
The CNMC found that Booking.com imposed unfair conditions on hotels and attempted to exclude other online travel portals, exploiting the economic dependence of hotels on its platform.
The final decision on the fine is expected in July, and if confirmed, Booking.com plans to challenge it through the appeals process.
3) Tech Leaders Fled San Francisco During the Pandemic. Now, They’re Coming Back.
Despite criticism of San Francisco's government and high living costs during the pandemic, the city is experiencing a tech revival, particularly in artificial intelligence. Silicon Valley leaders are returning to the city, investing in local politics and pushing for startups to return to the Bay Area and resume in-office work.
San Francisco's startup ecosystem has remained resilient, with investment dropping less compared to other tech hubs like Austin, Los Angeles, and Miami.
Tech founders and investors are returning to San Francisco despite its issues with homelessness and drug overdoses, driven by the city's tech resurgence and proximity to institutions like Stanford University.
Examples include the co-founders of fintech startup Brex returning to San Francisco after stints in Los Angeles, New York City, and Miami, and Howie Liu, CEO of Airtable, spending more time in San Francisco to meet with sales customers after relocating to Los Angeles during the pandemic.
4) Japan’s Nikkei 225 index eclipses record high after 34 years
Japan's Nikkei 225 index hit an all-time high last week, surpassing its 1989 record during a powerful rally fueled by chip-related stocks.
Closing above 39,000 points for the first time, the Nikkei's 2024 surge reflects a 17.5% rise since the year's start, making it the world's top-performing major index.
Strong US chipmaker Nvidia earnings contributed to investor confidence, with traders celebrating the milestone across Tokyo.
Japan's broader Topix index is also nearing its 1989 peak, with analysts predicting further gains for both indices by year-end.
Corporate earnings growth and governance reforms have bolstered Japan's market resurgence, attracting domestic and foreign investors alike.
Tool of the Week 📊
Last week we talked about DBT. If you missed it, check it out here -
This week, We are going to look at - Airflow
Apache Airflow provides a flexible and scalable platform for orchestrating complex workflows, making it a popular choice for data engineering, ETL (Extract, Transform, Load) pipelines, machine learning workflows, and other automation tasks in various industries.
The main characteristic of Airflow workflows is that all workflows are defined in Python code. Airflow can help with the following tasks precisely -
Manage scheduling and running jobs and data pipelines
Ensures jobs are ordered correctly based on dependencies
Manage the allocation of scarce resources
Provides mechanisms for tracking the state of jobs and recovering from failure
Basic Airflow concepts
Task: a defined unit of work (these are called operators in Airflow)
Task instance: an individual run of a single task. Task instances also have an indicative state, which could be “running”, “success”, “failed”, “skipped”, “up for retry”, etc.
DAG: Directed acyclic graph, a set of tasks with explicit execution order, beginning, and end
DAG run: individual execution/run of a DAG
Architecture Overview
A workflow is represented as a DAG (a Directed Acyclic Graph), and contains individual pieces of work called Tasks, arranged with dependencies and data flows taken into account.
A DAG specifies the dependencies between tasks, which defines the order in which to execute the tasks. Tasks describe what to do, be it fetching data, running analysis, triggering other systems, or more.
Who are the users of Airflow ?
The platform was created by a data engineer — namely, Maxime Beauchemin — for data engineers. No wonder, they represent over 54 percent of Apache Airflow active users.
Want to get started with Airflow? Check out this webinar for beginners.
Next time we will talk about a tool I am the most curious about this year
Until next edition,
Raghunandan 🎯
& If it’s your first time here, TheWeekendFreelancer currently has 5 ongoing series - Tools 🛠️, Maths 📈, Domain 🌐, Trends 📻 & My Notes 📝. Have fun reading!
P.S. - “The Weekend Freelancer” is a reader backed publication. Share this newsletter with your friends and relatives & consider becoming a free or paid member of this newsletter. Every subscription gives me an extra ounce of motivation to keep going 💪!
You can also support my work here. Cheers!