Reality vs. Hype: How AI & Human Activity Combine
AI-driven startups across the world weigh in on the impact of the AI revolution on businesses & industries
By: Denis Barrier & Bruno Delahaye
ChatGPT has dominated discourse amongst professionals and individuals alike — bringing AI to the masses for the first time. Is it hype or a real new tech revolution at hand? Will we soon see a long lasting impact on industries and society reverberate from today’s headlines?
The tech behind ChatGPT, and similar AI-tools or platforms, is not new. While Microsoft-backed OpenAI may have been the first to bring it mainstream, many of big tech’s big names have similar large language models (LLMs) (e.g., Google’s BERT and BARD or Facebook’s RoBERTa) and companies of all sizes have been leveraging AI (generative or not), machine learning, deep learning and natural language processing (NLP) for over a decade.
However, up until now, the mainstream use of AI was drastically different — particularly for white-collar vs. blue-collar work. With the former, AI primarily enabled efficiency in the human-dependent workplace by embedding in things like search (google), CRM (customer acquisition, cross-selling and upselling, churn) or supply chain (stock forecasting, or optimization). With the latter, AI promised a more profound impact as a potential replacement for more manual roles with physical robots and machinery such as Automated Guided Vehicles (AGVs) in warehouses or Articulated Robotic Arms in manufacturing.
More recently, the performance of AI has increasingly amazed the world with algorithms like AlphaZero (developed by Google’s DeepMind in 2017) mastering the games of chess, shogi and Go in the matter of hours. Boston Dynamics has also pushed the boundaries with highly advanced and agile robots completing complex movements and tasks previously thought impossible. But now, generative AI solutions are taking it to a whole new level by displaying human-like behavior in an almost unconstrained environment — be it through text (like ChatGPT) or images (like Dall-E).
In some respect, AI is displaying the capacity to surpass humans by outperforming on specialty exams (e.g., the Bar, Master Sommelier, US medical licensing exam, microbiology) — while comprehending and translating across 95 different languages. As human and AI interactions become more and more indistinguishable, seemingly inching closer to Artificial General Intelligence (AGI), today’s world is trying to grapple with the potential impact of what is the first true AI revolution on all types of work across industries — from automated deliveries to writing code, diagnosing patients and beyond.
At Cathay Innovation, we’ve been investing in AI-driven startups across Europe, North America and Asia since launching our first fund back in 2015. But rather than taking it from us — we asked the founders and startups within our portfolio, who have been developing cutting-edge AI solutions in various sectors, to share their perspectives on how they see the AI revolution unfolding. How will AI disrupt their industry? How will ChatGPT impact their business?
Is the impact real or hype?
Healthcare & Biotech
Thomas Clozel, Co-founder & CEO, Owkin | France & US
Owkin is a French-American AI biotechnology company that uses AI to find the right treatment for every patient. It was co-founded by Thomas Clozel MD, a clinical research doctor and former assistant professor in clinical onco-hematology, and Gilles Wainrib, a pioneer in the field of machine learning in biology.
AI will disrupt healthcare and biotech by augmenting the probability of success of drug programs and ushering in a long overdue promise to patients — precision medicine.
AI will help both on the drug discovery and development side. With human researchers + LLMs, research will be augmented with new data while AI (through federated learning) will enable a privacy-first and collaborative environment that furthers medical innovation — e.g., discoverving new biomarkers with more causality or multimodal treatments with new AI, pathology and omics-based technologies. The output will be new targets matched with new subgroups of patients of high-value as well as new hypotheses on disease mechanisms. This will all contribute to more precise and effective drug discovery and treatment strategies within diagnostic tools.
ChatGPT is an important milestone as it showcases the versatile capabilities of LLMs in generating human-like text across a wide range of tasks and use cases on an unprecedented scale. It already impacts many of our operations, from software engineering to marketing, by accelerating the production of relevant content.
There’s no doubt that LLMs will play an increasing role in scientific research and that we will see the emergence of LLM-based “smart agents” helping scientists generate and validate hypotheses. Current LLMs, with their tendency to “hallucinate”, are still too immature for scientific research. But “augmented LLMs” with extra capabilities (e.g., searching databases or running computations) represent a promising direction towards smart agents supporting scientists — particularly in biomedical research and drug discovery.
Eugene You, President & CEO, Lumi (Aqara) | China
Lumi develops smart home appliances and IoT solutions, including through its Aqara brand, providing customers with comprehensive smart home products and solutions that are beautifully designed, affordable and easy to use.
AI has a significant impact on the smart home industry, bringing a more intelligent, convenient, comfortable and safe home experience. For example…
Natural language understanding (NLU) technology enables users to interact with smart home products and services in a more natural way and improves convenience.
AI recognition technology provides more intelligent security for smart home systems, such as stranger recognition, fall detection or fire detection.
Data intelligence can be used to analyze large amounts of time-series data from smart hardware and sensors, learning user behavioral habits and preferences to provide personalized smart experiences.
The technology and product paradigm innovation brought about by LLMs (like ChatGPT) plays an important role. First, it can solve the limitations of traditional voice interactions with more accurate understanding and generation of natural language. More human, natural and convenient interactions enhance user satisfaction and stickiness to smart home products.
Second, by analyzing smart hardware and sensor data, LLMs can provide seamless personalized intelligent experiences and improve home energy efficiency. For example, it can proactively adjust room temperature or analyze data on electricity, water or gas to provide accurate energy-saving recommendations. Overall, LLMs can bring many innovative development opportunities to the smart home industry — which will improve user experience and quality of life at home.
Ari Tuchman, Co-founder & CEO, Quantifind | US
Quantifind is a US-based provider of AI-powered financial crimes risk management solutions serving Tier 1 banks, Fortune 50 companies and the Department of Defense. It was founded in 2009 by Ari Tuchman and John Stockton who met as fellow quantum physicists with a mutual passion for extracting signals from data.
AI is a natural evolution in disruptive technologies in fintech — providing a step function increase in what processes can be automated. With robotic process automation (RPA) — simple, static rule-based workflows can be replaced digitally, but with AI, complex and dynamic workflows in messy data situations can be replaced with automation. AI solutions can automate tasks ranging from personal recommendations of financial products to risk assessment for anti-money laundering (AML) and “know your customer” (KYC) checks. This degree of automation drastically reduces compliance and infrastructure costs for established banks and lowers the bar for new financial services to get up and running.
In AML/KYC applications, AI allows compliance teams to define risks, efficiently identify risky clients and counterparties, reduce fraud, and critically, reduce the hurdle for good parties to transact. These efficiencies are vital for supporting exploding volumes in digital transactions, and enabling the migration to interconnected Banking-as-a-Service offerings that touch all aspects of the digital economy. As touch points with customers in digital finance become less personal, AI-based trust determination of the customer becomes more vital.
ChatGPT continues the disruption in expanding what can be automated. Now, customer interactions, SAR filings, risk profiles and even appraisals can be automated. With careful prompts and guardrails, these can be less biased than manually generated content. ChatGPT applications currently represent use cases where medium accuracy suffices but can also provide powerful training data for high accuracy applications such as fraud, money laundering and threat detection. Depending on the application, ChatGPT will be both enduring and feature table-stakes.
John Brisco, Co-Founder & CEO, Coherent | SEA & US
Coherent is a global SaaS company working with some of the largest and most trusted brands in financial services, insurance and beyond. Its core technology, Spark, is a no-code SaaS platform that converts spreadsheets into enterprise-grade code.
The impact is very real and will be a major contributor to ongoing technology and business innovation. What’s fascinating is the accelerated sophistication of ChatGPT in such a short time frame. It will lead to smarter and faster research and recommendations along with new ways to train humans on novel concepts or analyze patterns and problems faced across businesses and society. The possibilities are truly remarkable.
As a data and logic heavy industry, it’s natural that financial institutions want to unlock more insight, opportunity and automation. The AI evolution in financial services will come in stages: from conversational enhancements to processes and then decision making or modeling. The problem statements and use cases will differ, but the opportunity and benefits will be potentially massive.
I’ll caveat this by saying that human oversight of AI activities or decisions will still be a critical control component. It’s not a “machine taking over” scenario. It’s a movement that uses human intuition and experience to maximize AI outcomes. There will likely be a massive focus on regulating AI as human instinct requires auditability on how decisions are made. Nevertheless, I’m a believer in the opportunity ChatGPT and others are creating — now it’s about how to integrate into systemic human and business behaviors.
Tanguy Touffut, Co-Founder & CEO, Descartes Underwriting | France
Founded by a team of insurance veterans and climate scientists, Descartes Underwriting operates at a global scale, offering data-driven parametric insurance against climate risk.
AI’s primary impact on the insurance industry will be in helping with simple and time-consuming tasks. For example…
Pricing and underwriting: AI techniques can be used to sort through complex and heavy weather data to better understand climate change and its consequences in terms of natural disaster frequency and intensity — one of our key focuses at Descartes.
Claim handling: computer vision is a game changer for claim assessment in the face of climate change. It can help improve the economics of non-life insurance through lower claim handling costs. It can also drastically reduce the time required between the declaration of loss and the claim payment — 550 days on average in the US (corporate segment) to just a few days.
Legal: extracting and processing data from complex documents is a big challenge for insurance companies and the gains can be massive. AI will increase underwriters’ analysis capabilities, e.g. checking clauses, spotting data inconsistencies etc.
While it’s still too early to assess its full potential, ChatGPT is clearly impressive but it still has limitations and requires “somebody” to double check the work. At this stage, for us operating in the non-life insurance corporate segment, ChatGPT, Google Bard and other similar technologies may contribute to efficiency gains — but it’s more of a tool for certain tasks than a complete disruption.
Mei Siauw, Co-founder & CEO, LeadIQ | SEA & US
LeadIQ is an account-based prospecting SaaS platform that simplifies lead management and sales prospecting. It was co-founded by former Oracle product manager Mei Siauw to optimize the entire end-to-end sales cycle.
AI will completely change the way we work by performing many of the manual tasks we do today, and greatly improving productivity. For example, in the sales and marketing industry, we’re already seeing AI writing first draft versions of emails, blogs and other forms of copy. But human intervention is still needed where it lacks judgment, empathy and intuition.
ChatGPT is not hype. Our team has been building an AI solution capable of writing sales emails based on context inputs, which our users are already using to augment work and increase productivity. With the speed of AI innovation, teams that evolve quickly to adapt to AI will be the ones most likely to win. Very soon, most humans will be working together alongside AI, with AI augmenting and complementing humans.
Arvid Wang, Chairman & CEO, Laiye | China
Laiye is a leading Robotic Process Automation (RPA) and Artificial Intelligence (AI) provider with a global presence across Europe, the Middle East and Africa (EMEA), Latin America, and Asia Pacific.
In the future, we will only have truly intelligent software — no longer costly and complex. The next generation AI-powered software will be so intuitive and effective that it could make existing category leaders, who don’t fully embrace AI, obsolete. For example, Customer Service CRMs face a direct existential threat if chatbots of human-level intelligence can be developed and maintained with ease.
Amara’s Law says we tend to overestimate the effect of a new technology in the short run and underestimate it in the long run. I firmly believe that ChatGPT and LLMs will have a significant impact on Laiye’s product offerings and the operating model of all organizations.
For example, LLMs will elevate our consumer-facing companion bot to be even more personalized and engaging while significantly reducing the development and model training costs for our B2B products. It will also allow us to unify all our automation and AI tools into one integrated platform with a natural language interface for all use cases of a digital workforce. Many of the knowledge-based roles, from marketing, customer service, product design to even software development, are likely to become transformed or augmented.
Organizations with a digital worker first mindset will achieve structural advantage in productivity against competition. And there will be hundreds of millions of AI citizen developers equipped with LLM-powered tools to usher in this renaissance of work.
What’s clear? The impact of AI on various industries is rapidly evolving. While ChatGPT and other AI-technologies still have limitations, most believe that given the mass gains in efficiency, accuracy and productivity, the not-so-distant future workforce will combine AI+humans. What’s not clear? What the full impact of AI will be — from an additive tool that augments or assists specialists vs. completely replacing some jobs.
We believe the mainstream adoption of ChatGPT marks the arrival of a new generation of AI-driven platforms that will become pervasive in industries. Today, we’re at the beginning of a major inflection point — where advanced technology, industry and society finally meet and will realize its potential on a massive scale exponentially larger than before. This is both an imminent threat to all businesses and a great opportunity for those that will be fast enough to harness the power of AI.
Building in the space? Do reach out, we’d love to hear from you!