Humans Less Expensive Than AI

☕️ Greetings, and welcome to my daily dive into the Generative AI landscape.

Check out my profile to see where I'm coming from, and swing by INVENEW to see what I'm up to these days. Also, I've got this new blog, ReROAR, where I write about AI, the future of work and living, cool solopreneurship ideas, and the must-have skills in our AI-driven future. 

In today’s issue:

• Databricks Announces the Industry’s First Generative AI Engineer Learning Pathway and Certification

• Chatbots are developing an understanding of the world, scientists claim

• Generative AI is expensive, and Microsoft needs to cut costs

• A Guide to Mastering Large Language Models

• Google GenAI Comes to Chrome

• And more

🔔 Please forward this newsletter to your friends and team members and invite them to join. This will help me grow my reach. Thanks, Qamar.

🗞️ Humans Less Expensive Than AI

A recent MIT study offers a reassuring outlook on the impact of AI on job automation, countering the prevalent fear of robots replacing human labor. The study, "Beyond AI Exposure," examines the feasibility of substituting AI for human tasks in various U.S. occupations, with a focus on computer vision-related tasks. It concludes that currently, only a small fraction of jobs can be cost-effectively automated using AI. Key factors include the high costs of AI implementation and maintenance, as well as its limitations in tasks requiring intuition, emotional intelligence, and critical thinking. Despite some industries being ripe for increased automation, the study suggests that human labor retains an edge in many areas, with AI's role in job displacement appearing less imminent than often feared.

Key Points

• MIT Study's Core Finding: Most jobs are still too costly to automate with AI, especially those requiring computer vision.

• Economic Feasibility: Presently, only 23% of wages in certain jobs can be economically replaced by AI. Full automation of vision tasks won't be economically viable until at least 2026.

• Long-Term Predictions: By 2042, there will still be tasks where human labor is more advantageous despite AI advancements.

• Limited Current Automation: As of now, only 3% of visually assisted tasks are economically feasible to automate, potentially rising to 40% by 2030.

• Human Advantage: AI struggles with intuition and emotional intelligence, areas where humans excel, making certain human skills irreplaceable.

• Industry Impact: Some sectors like banking and healthcare are more likely to see increased automation due to the nature of repetitive tasks.

• Overall Perspective: The fear of AI leading to widespread job loss is currently overblown, with human labor maintaining significant advantages in many fields.

Reference: AI is too expensive to replace humans in jobs right now, MIT study finds

🗞️ TRENDS

Recent advancements in artificial intelligence, particularly in the realm of large language models (LLMs), suggest a deeper level of understanding and creativity than previously acknowledged. Princeton's Sanjeev Arora and Google DeepMind's Anirudh Goyal propose that as LLMs like GPT-4 grow in size, they demonstrate capabilities beyond their initial training data. This hypothesis emerged from their observations of LLMs solving complex problems and seemingly generating original content. A study involving GPT-4, although not yet peer-reviewed, showcased its ability to integrate multiple skills into a coherent response, hinting at a form of understanding and creative synthesis.

Key Points:

1. Growing Intelligence in LLMs: Research indicates that as LLMs expand, they display outputs that likely weren't part of their original training, suggesting an evolving understanding of the world.

2. Surprising Abilities: LLMs have shown proficiency in complex tasks like mathematical problem-solving and inferring human thoughts, raising questions about the origins of these skills.

3. Study on GPT-4: A study, yet to be peer-reviewed, involved GPT-4 demonstrating an integration of multiple skills, including bias, metaphor, and physics, in a creative manner.

4. Evidence of Creativity and Understanding: The outputs from LLMs, such as GPT-4, are seen as more than mere mimicry of training data, indicating a form of compositional generalization and creativity.

5. Perspectives from Experts: External experts like Microsoft’s Sébastien Bubeck recognize these developments as significant, highlighting the essence of creativity in LLMs' ability to combine previously unlinked concepts.

Reference: futurism.com (2024). Chatbots are developing an understanding of the world, scientists claim

🗞️ IMPACT (Economy, Workforce, Culture, Life)

In 2024, governments globally, especially in emerging economies, are undergoing a profound transformation by embracing AI. This shift extends from generative AI applications to comprehensive community upskilling programs, revolutionizing governance. While AI enhances efficiency, it necessitates a populace with AI literacy. Striking a balance between technological progress and human development is pivotal. Governments are focusing on strategies that merge AI solutions with educational and training programs, ensuring both technology optimization and human capacity building. Generative AI takes center stage, streamlining operations and enhancing citizen services. Case studies from the German Federal Foreign Office and Bhutan exemplify AI's strategic utilization. NGOs are also harnessing AI's power, emphasizing ethical, community-driven approaches. This era marks technology's role in creating a better-connected, accessible world.

Reference: A Turning Point for AI in Government in 2024 – From Generative AI to Community Upskilling

🗞️ OPINION (Opinion, Analysis, Reviews, Ideas)

ChatGPT, OpenAI's AI chatbot, has significantly impacted the tech world and beyond. Initially a productivity tool, it's now a staple for over 92% of Fortune 500 companies. However, recent leadership turmoil at OpenAI, including the temporary ousting of CEO Sam Altman, has sparked industry concerns. This situation could benefit rivals like Meta and other AI startups, drawing attention and funding. Despite controversies, ChatGPT's influence is undeniable with 100 million weekly users and continual investments. OpenAI's first developer conference highlighted major updates, including GPT-4 Turbo and a multimodal API. GPT-4 remains largely premium, but accessible via Bing Chat in various web browsers. Additionally, OpenAI has integrated internet connectivity and DALL-E 3 into ChatGPT, enhancing its capabilities.

Reference: TechCrunch.com (2024). ChatGPT: Everything you need to know about the AI-powered chatbot 

🗞️ LEARNING (Tools, Frameworks, Skills, Guides, Research)

Databricks introduces the industry's first Generative AI Engineer learning pathway and certification to address the growing demand for generative AI expertise. In a swiftly evolving tech landscape, upskilling is imperative. The program offers self-paced and instructor-led courses, culminating in the Generative AI Engineer Associate Certification exam. With the average half-life of tech skills shrinking, this initiative equips professionals with vital knowledge. The Generative AI Fundamentals course, launched in 2023, complements this pathway. Join the journey to master generative AI and shape its future.

Reference: Databricks Announces the Industry’s First Generative AI Engineer Learning Pathway and Certification

🗞️ BUSINESS (Use Cases, Industry spotlight)

The U.S. FTC is currently scrutinizing prominent generative AI collaborations involving tech giants such as Microsoft, Google, and Amazon. These investigations hold the potential to reshape the AI landscape. Simultaneously, the AI industry is witnessing a shift away from building models from the ground up, with approximately 95% of AI spending directed towards inference over training. This transformation highlights the growing significance of APIs, offered by entities like OpenAI and Anthropic, and prompts inquiries into how companies can differentiate themselves competitively, particularly through effective data integration in AI development.

Reference: fortune.com (2024). AI’s big shift from ‘model-forward’ innovation to ‘product-forward’

🗞️ LATEST FROM THE WEB

Google GenAI Comes to Chrome: Google has unveiled experimental AI features in Chrome M121, aiming to enhance browsing with personalized, efficient experiences. Notably, the Tab Organizer automates tab grouping and suggests names, streamlining multitasking. Users can create custom Chrome themes through AI image generation and access pre-made options. An AI writing assistant aids content creation with suggestions for various tasks. These features are accessible through Chrome settings, though temporarily disabled for enterprise and educational accounts. Google plans to expand AI integration in Chrome throughout 2024, with a focus on improved browsing speed and convenience, including the introduction of the Gemini AI model. Read more thenewstack.io.

Pegasystems launches generative AI-based assistant for enterprises: Organizational data fragmentation poses challenges for employees and customers alike, making information retrieval cumbersome. Pegasystems addresses this with "Knowledge Buddy," a solution that condenses scattered data into context-aware guidance, seamlessly integrating it into operational portals. This versatile tool extends to customer-facing platforms, reducing the need for call centers. Notably, Knowledge Buddy offers enterprise-level features such as content auditability and security. Pegasystems targets existing clients, aiming to enhance their workflow applications. In a growing landscape of generative AI, concerns about accuracy and transparency persist, but Knowledge Buddy offers promise in tackling information disarray. Read more at computerworld.com.

Generative AI is expensive, and Microsoft needs to cut costs: Microsoft is strategically investing in smaller, more efficient AI models through its GenAI team, led by Misha Bilenko, aiming to replicate the quality of large models while cutting computing costs. They've reassigned top AI developers to this initiative, including Sebastien Bubeck. This cost-efficiency drive responds to the challenge of scaling AI technology, where expenses can skyrocket. Microsoft faced losses on Github Copilot, with some users costing up to $80 per month. The broader industry also seeks alternatives to costly AI chips like Nvidia's, exploring efficiency improvements without sacrificing quality. Read more at the-decoder.com.

How Perplexity’s Online LLM Was Inspired by FreshLLMs Paper: Perplexity AI's Copilot, unveiled at AWS re:Invent 2023, combines generative AI and search, addressing LLM challenges: outdated data and hallucinations. Web search APIs and retrieval augmented generation enhance responses. Inspired by "FreshLLMs," Copilot injects context from search results and supports few-shot prompting. FreshQA benchmark reveals LLM limitations. Online models, pplx-7b-online and pplx-70b-online, offer real-time data access via API. Regular fine-tuning ensures quality, and an API credit is available for Pro subscribers. Upcoming article: a tutorial on building applications with Perplexity AI's API. Read more at thenewstack.io.

A Guide to Mastering Large Language Models: Large Language Models (LLMs) have surged in popularity, transforming natural language processing and AI across industries. These models, like GPT-3, are trained on vast text corpora and revolutionize tasks from chatbots to creative writing aids. LLMs represent a paradigm shift, powered by internet-scale data, massive model sizes, and self-supervised learning. To effectively leverage LLMs, understanding key concepts like prompting, embeddings, attention, and retrieval is crucial. Architectural patterns like text generation pipelines, search and retrieval systems, multi-task learning, and hybrid AI approaches enable practical applications. Developing skills in prompt engineering, orchestration frameworks, evaluation, monitoring, and embracing multimodal applications ensures proficiency in harnessing LLMs' potential. Read more at unite.ai.

In partnership with Amazon Web Services

🔔 Architecting continuous Observability from development to production

Save your spot:  February 14, 2024 | 11:00 AM PDT | 2:00 PM EDT

Join this webinar to explore how to establish effective Observability in dynamic, continuously deployed, and decoupled systems. You’ll discover where and what to instrument across your SDLC and how to leverage clear insights in today's evolving tech environment.

You will learn how to:

▶ Choose an optimal data type and service to monitor the progression of your applications throughout the SDLC.
▶ Enhance your ability to store, query, and visualize system telemetry for more effective insights into your systems.
▶ Enable transparency and address issues proactively before they impact operations to systematically enhance value delivery and resiliency.
▶ Use tools from AWS and the broader DevOps tool landscape using AWS Marketplace to establish a comprehensive Observability practice.

Your Feedback

I want this newsletter to be valuable to you so if there's anything on your mind—praises, critiques, or just a hello—please drop me a note. You can hit reply or shoot me a message directly at my email address: [email protected].

Join my community by subscribing to my newsletter below: