End of 2020: Milestones and Momentum in AI

2020 has been a crazy year, as we all know. For me and my company, Conversica, it has been the year of Revenue Diversity….trying to take a business of singles (single use-case in sales; single buyer, in marketing; single segment, SMB; single industry, auto; single sales G2M, inside sales; single region, north america; narrow AI architecture, ML/DL) and begin to diversify into other areas that could potentially increase revenue.

As visionaries of AI, this last area – this AI architecture of machine learning (ML) and deep learning (DL) is where we have to look for product-led-growth. Here we have to begin to lay the groundwork for the future of AI applications for sales teams and more. At Conversica we tried to apply OpenAI’s GPT-2, then GPT-3, GPT-Neo, Blenderbot, BERT, and a whole host of various models. We have been trying to disrupt the chatbot space first. Then, leveraging this breakthrough technology into our core in email and SMS. Then across all channels of communication including messaging (WhatsApp, WeChat, Lime, imessage, FB messenger, Twitter, Linkedin, Slack, etc), and voice. “Conversational AI” will eventually become a household phrase.

As we continue to define our place in AI, I am witnessing a period of remarkable progress and excitement in the field of artificial intelligence, especially in the development and application of Large Language Models (LLMs). Here are several notable events and trends in 2020 that are shaping the AI landscape:

  1. GPT-3 Release by OpenAI (June 2020): OpenAI released GPT-3, the third generation of its Generative Pre-trained Transformer model, in June 2020. It also licensed it to Microsoft. With 175 billion parameters, GPT-3 set a new benchmark for language models, demonstrating unprecedented capabilities in natural language understanding and generation. The model’s ability to perform various tasks with minimal fine-tuning showcased the power of scaling up language models, capturing the attention of researchers, developers, and businesses worldwide.
  2. Widespread Adoption and Integration of GPT-3: By the end of 2020, GPT-3 is now being integrated into numerous applications and platforms (see a few below). Companies and developers leveraging its capabilities to build innovative solutions in areas such as customer service, content creation, coding assistance, and more. The model’s versatility and performance has fueled a surge in interest and experimentation, driving the adoption of AI-powered tools across industries.
  3. AI for COVID-19 Response: The global COVID-19 pandemic accelerated the adoption of AI technologies to address various challenges. AI and LLMs played crucial roles this year in areas such as epidemiological modeling, vaccine research, medical diagnosis, and information dissemination. Conversica and other AI companies focused on deploying intelligent solutions to support businesses in navigating the crisis.
  4. Advancements in Multimodal AI: Researchers began exploring multimodal AI, which combines text, image, and audio data to create more comprehensive and context-aware models. This trend aimed to enhance the capabilities of AI systems by enabling them to understand and generate content across different modalities, paving the way for more sophisticated and interactive AI applications.
  5. Ethical AI and Responsible AI Development: As AI technologies advanced, there was growing emphasis on ethical AI and responsible development practices. Concerns around bias, transparency, and accountability led to increased efforts to establish guidelines and frameworks for ethical AI. Industry leaders, including Conversica, prioritized the development of fair and transparent AI systems, engaging in discussions on the societal impact of AI.
  6. Transformers in Research and Industry: The transformer architecture continues to gain traction in both research and industry. Beyond OpenAI, other organizations and academic institutions were actively working on improving transformer models, exploring new architectures, and pushing the boundaries of what LLMs could achieve (see below). These efforts aimed to enhance the efficiency, interpretability, and robustness of language models.
  7. Collaboration and Open Science: The AI community emphasized collaboration and open science, with researchers sharing their findings and models openly to accelerate progress. Initiatives like Hugging Face’s Transformers library made it easier for developers to access and experiment with state-of-the-art models, fostering innovation and knowledge exchange.
  8. Emergence of Specialized LLMs: In addition to general-purpose LLMs like GPT-3, there was a growing interest in developing specialized models tailored to specific domains or tasks. These models aimed to deliver higher performance and accuracy in areas such as legal document analysis, scientific research, and financial forecasting.

At Conversica, we are at the forefront of these developments too, steering the company to harness the power of LLMs and AI technologies to drive innovation and deliver value to our clients. The end of 2020 marks a pivotal moment in our AI journey, setting the stage for even more groundbreaking advancements in the years to come.

Transformative Applications and Platforms Integrating GPT-3 by the End of 2020

By the end of this year, GPT-3 has been integrated into a variety of applications and platforms that demonstrated its transformative potential across multiple domains. Here are some of the most notable ones:

  1. Copy.ai: Copy.ai utilized GPT-3 to help users generate marketing copy, social media posts, and other forms of written content. By leveraging GPT-3’s natural language generation capabilities, Copy.ai enabled businesses to streamline their content creation processes and improve the quality and consistency of their marketing materials.
  2. Replika: Replika, an AI chatbot designed to provide companionship and mental health support, integrated GPT-3 to enhance its conversational abilities. This allowed the chatbot to engage in more meaningful and contextually aware conversations with users, providing better emotional support and personalized interactions.
  3. AI Dungeon: AI Dungeon, an interactive text-based adventure game, used GPT-3 to generate dynamic and complex narratives based on player input. This integration created a highly immersive and adaptive gaming experience, where players could explore an endless array of storylines and scenarios.
  4. Kuki: Kuki (formerly Mitsuku), a conversational AI chatbot, leveraged GPT-3 to improve its natural language understanding and response generation. This enhancement allowed Kuki to engage in more nuanced and sophisticated conversations with users, making it a more effective and engaging virtual companion.
  5. Viable: Viable used GPT-3 to analyze customer feedback from various sources, such as surveys, support tickets, and social media. By processing and summarizing this data, Viable provided businesses with actionable insights and trends, helping them improve customer satisfaction and product development.
  6. Fable Studio: Fable Studio integrated GPT-3 into its AI characters, enabling them to have more realistic and engaging conversations with users. This innovation was particularly transformative in the realm of interactive storytelling and virtual reality, where lifelike AI characters enhanced the overall user experience.
  7. Debuild: Debuild utilized GPT-3 to generate code from natural language descriptions. This platform allowed developers to describe the functionality they needed in plain English, and GPT-3 would generate the corresponding code, significantly speeding up the development process and making coding more accessible to non-programmers.
  8. Chatbots and Customer Service: Many companies integrated GPT-3 into their customer service chatbots to improve automated support interactions. GPT-3’s ability to understand and generate natural language responses enabled these chatbots to handle more complex queries, provide accurate information, and deliver a more human-like customer experience.
  9. Jasper.ai (formerly Jarvis): Jasper, a platform for writing assistance, used GPT-3 to help users with various writing tasks, including drafting blog posts, emails, and reports. This tool empowered writers by providing suggestions, improving the flow of their content, and saving time on tedious writing tasks.
  10. Email and Messaging Automation: Platforms like Compose.ai integrated GPT-3 to assist with email and messaging automation. These tools helped users draft personalized and contextually relevant emails, saving time and enhancing communication efficiency.

These applications and platforms showcased GPT-3’s versatility and transformative impact across different industries, demonstrating how advanced language models could revolutionize content creation, customer service, interactive entertainment, and more.

Transformative Industry Research by the End of 2020

This year, several organizations are emerging as THE notable players in the development of Large Language Models (LLMs), alongside OpenAI. These providers are making significant contributions to the advancement of natural language processing (NLP) and AI. Here are some of the key emerging LLM providers:

  1. Google AI / Google Research: Google continues to be a major force in AI research, building on their work with the BERT model. They were developing increasingly sophisticated models and techniques, such as T5 (Text-To-Text Transfer Transformer), which unified various NLP tasks under a single framework. Google’s contributions to the transformer architecture and NLP research were foundational to many advancements in the field.
  2. Facebook AI Research (FAIR): Facebook’s AI division, FAIR, was actively working on large-scale language models. Their work included the development of models like RoBERTa (Robustly optimized BERT approach) and other variations that focused on improving the robustness and performance of language models. FAIR’s research emphasized scalability and practical applications of LLMs in social media and content moderation.
  3. Microsoft Research: Microsoft, often in collaboration with OpenAI, was deeply involved in advancing LLM technologies. They provided significant computational resources and integrated LLMs into their products, such as the Azure AI platform, which offered various NLP services powered by these advanced models. Microsoft’s investment in AI research and infrastructure positioned them as a key player in the LLM space.
  4. Hugging Face: Hugging Face was rapidly becoming a pivotal organization in the NLP community with their Transformers library, which made state-of-the-art models like BERT, GPT-2, and T5 accessible to a wider audience of developers and researchers. They were fostering an open-source ecosystem that facilitated the development, training, and deployment of LLMs.
  5. DeepMind: DeepMind, a subsidiary of Alphabet, was known for its cutting-edge research in AI, including language models. Their work on models like AlphaFold (though primarily focused on protein folding) and other deep learning projects showcased their capabilities and contributions to the broader AI landscape.
  6. Alibaba DAMO Academy: Alibaba’s DAMO Academy was another notable player, particularly in the Asian market. They were developing large-scale AI models and applying them to various applications, including e-commerce, logistics, and customer service. Their work was influential in the global AI research community.
  7. Tencent AI Lab: Tencent’s AI Lab was engaged in research and development of LLMs, contributing to advancements in machine translation, content generation, and conversational AI. Their efforts were particularly impactful in the context of social media and gaming.
  8. Baidu Research: Baidu, often referred to as the “Google of China,” was investing heavily in AI research, including the development of large language models. Baidu’s focus on integrating AI into search engines, autonomous driving, and cloud services highlighted their role in the AI ecosystem.

These organizations are at the forefront of LLM development, contributing to the rapid evolution of AI technologies. Their collective efforts were driving significant advancements in natural language understanding, generation, and application, paving the way for the transformative AI era we see unfolding today.


Jim Kaskade

Jim Kaskade is a serial entrepreneur & enterprise software executive of over 36 years. He is the CEO of Conversica, a leader in Augmented Workforce solutions that help clients attract, acquire, and grow end-customers. He most recently successfully exited a PE-backed SaaS company, Janrain, in the digital identity security space. Prior to identity, he led a digital application business of over 7,000 people ($1B). Prior to that he led a big data & analytics business of over 1,000 ($250M). He was the CEO of a Big Data Cloud company ($50M); was an EIR at PARC (the Bell Labs of Silicon Valley) which resulted in a spinout of an AML AI company; led two separate private cloud software startups; founded of one of the most advanced digital video SaaS companies delivering online and wireless solutions to over 10,000 enterprises; and was involved with three semiconductor startups (two of which he founded, one of which he sold). He started his career engineering massively parallel processing datacenter applications. Jim has an Electrical and Computer Science Engineering degree from University of California, Santa Barbara, with an emphasis in semiconductor design and computer science; and an MBA from the University of San Diego with an emphasis in entrepreneurship and finance.