Wojciech Gryc shares a post on how GPT-3 impacts the user experience.
A great deal has been written about GPT-3 and its potential impact and hype on AI, machine learning, and data science. This post aims to look at the user experience around GPT-3 instead. Specifically, why do some people see GPT-3 as a magical innovation? What does this tell us about AI-driven products we don’t fully understand?
I’ve been exploring GPT-3 for the past few weeks and have been incredibly impressed with its ability to take my natural language prompts and generate helpful responses. More importantly, I’ve organized a few demo events and discussions about GPT-3 and have seen people play and interact with it.
The GPT-3 Click
When a user first logs into the GPT-3 Playground (i.e., demo interface), they are met with an empty text box and several modeling options. You’re expected to write, to chat, or input some text–what you say or do is up to you. It can be overwhelming in its simplicity.
There are sample text prompts to show you how to generate a Q&A session, or an English/French translation, or a story… What is critical, however, is that the entire interface is a text box–as the user, you simply provide text (i.e., the prompt) and ask GPT-3 to do the rest.
Key points include:
- New User habits
- Lessons for AI-driven Products
- Play and forgiveness, speed and scalability
Read the full article, User Experience and AI: the GPT-3 ‘click’, on 10MillionSteps.com.
It’s always interesting to take a look back while stepping forward. In this older post from Supriya Prakash Sen, the use of AI technology in the workplace was explored. How does it compare to today’s outlook?
The news has been awash with provocative articles about the future of jobs in our society. The exponentially advancing nature of Artificially Intelligent machines, after AlphaGo turned out to be a better Go player than any human – combined with the power of the collective mind, makes it an urgent question to debate. There seems to be almost no job or field of endeavor that cannot be disrupted – from routine and manual jobs to non-routine and cognitive jobs, all are now at risk of being replaced by intelligent machines.
Simple example- the other day I saw a conscious robotic arm in the pharmacy of the hospital, which is already dispensing medicine packages more accurately, efficiently and in a more space-saving way than any human could possibly do. Similarly – robotic arms that sort through waste at landfills are more productive, and also cheaper in the long run, albeit replacing work for humans who sort through garbage (sadly, often the first form of entrepreneurship for the disenfranchised). This raises the question, that maybe humans should let the work be done by machines after all- why fight it – we humans were meant for more higher pursuits anyway? Meanwhile robot bartenders are already employed in ships, see video clip: Robot Bartender on Cruise Ship.
Machine vs. Man was never a fair fight. From cameras and telescopes to ships and airplanes and drones, to the newest generation of “thinking” computers- there are hardly any jobs that machines cannot do better than humans.In fact, recent advances in technology and networked intelligence can lead to massive changes in entire societies, in the space of less than a generation. For just a small instance, look merely at what Fitbit can accomplish through scale and peer-pressure – rippling through an entire population, changing habits and behaviors in a relatively short period of time- and compare this with the impact a Personal Trainer can have with one client in a long set of focused one-on-one interactions.
Key points include:
- Extending human capability
- Universal basic income
- Virtual rewards replace money
Read the full article, The Power and the Fear – Artificial Intelligence and its impact on Jobs and Society, on LinkedIn.
The future of work, agriculture, education, and even relationships are all areas facing change thanks to AI technology. David Edelman extols the benefits of AI in this post.
The digital explosion, accelerated by Covid, has not made life on the front lines of sales and customer service any easier. In fact, when customers are able to do more research on their own, salespeople face tougher unanswered questions, and more of an inquisition about competitive differences, granular product details, or use cases they’ve never considered. Service reps have to handle the calls of customers facing challenges they could not resolve online, likely meaning customers who are more frustrated or who have very complex situations, often demanding special treatment or deeper investigation. And if they cannot work in a call center setting, getting help from colleagues or managers is simply more challenging logistically.
No matter what prognosticators say about AI automating away jobs, there will always be a need for front line roles (even if fewer people can handle many more calls) and AI can supercharge them by augmenting the capabilities available at the rock face of customer interaction. Reps will be more effective, and as their efficiency in “handling the difficult” goes up, they will become more scalable. The business cases are getting powerful.
Key points include:
- The new powers of augmentation
- It’s a brand issue
- But the tools are not enough
Read the full article, AI to the Rescue, as Call Centers Struggle
David Burnie shares a post from his company blog on how automation in the first step of claims processing can help streamline the process.
The First Notice of Loss (FNOL) – the first step in claims processing – is one of the most crucial customer touchpoints for an insurer. Yet, for most carriers, FNOL continues to be a lengthy, manual, call centre-based service requiring extensive data gathering. This process translates to high operational costs and cycle time and a less than satisfactory customer experience.
Providing a fast, streamlined and transparent claims intake process is no longer aspirational; it is table stakes for customers who expect nothing less from their interactions with all their service providers, including banking, retail, and entertainment.
Luckily for insurers, the intelligent automation landscape has advanced significantly over the last few years, enabling insurers to innovate rapidly and cost-effectively.
Traditionally, the key barriers to change for insurers included long-established processes that rely on legacy systems and a workforce under strain. A transformation roadmap for claims starts with re-imagining the end-to-end customer experience at each stage of the claim. Intelligent automation and artificial intelligence (AI) offer a proven pathway to produce a better claims service while leveraging core legacy systems. Intelligent automation brings systems, both legacy and new, into the process to create a seamless experience.
Key points include:
- Automating FNOL in auto insurance
- Automating FNOL in travel insurance
- Benefits of intelligent automation in insurance
Read the full post, How Automation Can Support First Notice of Loss (FNOL) Reports, on the Burniegroup.com.
David Edelman explains how having an AI strategy can provide a wide range of new variables that can be personalized to the customer journey, helping to speed and scale and make the concept of agile a reality.
Champion versus challenger has been the basis for finding lift and proving marketing value for decades. But we are way past the slower, high-cost operations needed to test direct mail pieces or even direct response TV. To find the right offer, message, or call center interaction on an increasingly personalized basis for digital channels requires multi-variate testing on a massive scale. Even the fastest “war rooms” for rapid test and learn, operations I’ve helped many companies build, suffer if you need analytic experts to create every test matrix, copywriters to draft every word for every variation, and operations managers to coordinate layers of simultaneous tests. And would you even be able to find all the talent you need at an affordable cost?
New optimization tools, powered by artificial intelligence, have already infiltrated the media buying process. And there is an overwhelming range of tools to turn first- and third-party data into targeting plans. But now we can use AI to optimize the personalization of the CONTENT of an interaction. Offer design, words, pictures, conversational interactions, tonality, and hundreds of other descriptors become data that can be measured, matched, and modelled.
Key points include:
- AI as a growth engine
- Enhancing the emotional connection
- Driving a rethink of operations
Read the full article, What’s your AI strategy for scaling Segment-of-One Marketing? on LinkedIn.
Tobias Baer draws attention to the danger of selective perception becoming the norm as the use of AI in online information and marketing limits the amount of information delivered.
There is a famous psychological experiment where participants intently watch a basketball game – but when asked afterwards about the gorilla that had danced around amidst the players, nobody has seen it. It’s the literal textbook example of selective perception – in this experiment, participants were tasked with counting the number of passes between the players and as they focused all their attention on the ball, their minds completely disregarded everything else going on on the court.
If you think of selective perception as a curtain that is partially drawn on our minds, thus narrowing our window into the world, AI is pulling more curtains from every side, leaving only a dwindling beam of light. If we don’t actively manage this and make sure we get enough exposure to mental sunlight, we risk making increasingly poor decisions and falling prey to manipulation by marketers. In the following, I will quickly describe how selective perception affects our beliefs and actions before reviewing some of the recent innovations in how AI is used that worry me for what they could do to our perception.
Our own selective perception is technically necessary but also a key way how our personality manifests itself. You all will have met anxious people who seem to always only see the risks of a proposal, or helpless optimists who seem to be blissfully blind to any risks or downsides.
Key points include:
- Facebook’s acquisition of Kustomer
- GPT-3, a language prediction model
- Side-tracked cognitive processes
Read the full article, How AI closes the curtain on human perception, on LinkedIn.
Christophe De Greift explains why now is the time to plan your analytics transformation, why you should, and the first step to take.
Artificial intelligence is a relatively young discipline in companies and in constant evolution. However, the experience of pioneering companies in various sectors and continents confirms their high value generation when accompanied by a true transformation of the company. Those who have not yet internalized their analytical transformation plan should start now to be able to arrive on time, as I explain below.
Being analytical is increasingly necessary to stay competitive
Many important business decisions are better when supported by data. Neuroscience has recently confirmed what common sense has always allowed us to understand: man can be irrational, biased and even blind. The machine is not 100% reliable either; the key is to precisely define the role of man and machine for each decision, as explained in a previous article . In very repetitive problems such as forecasting the demand for mass consumer products in retail stores, the most advanced and successful companies limit human intervention to exceptional cases. In more sensitive problems such as medical diagnoses, the radiologist is assisted by artificial intelligence.
Across all sectors, the competitiveness gap between analytics pioneers and the others is growing, threatening the sustainability of the latter. Those who have succeeded in analytics redouble their efforts and investments to become even more analytical, creating the virtuous circle explained in a recent MIT and BCG article.
Being analytical requires a transformation at the people, organization, processes and technology level.
Key points include:
- Data as an opportunity
- Predictive maintenance
Read the full article, 3 reasons to plan your Analytics Transformation now, on christophedegreift.com.
Umbrex is pleased to welcome Abhinav Chandra. Abhinav is a seasoned Senior Leader with deep expertise in Retail/Consumer and Technology across a wide range of functional areas. Abhinav was an Associate Partner in McKinsey’s Retail and Operations practices where he served multiple clients on topics including omni-channel supply chain strategy, marketing and growth strategy, store operations and private label design and sourcing to name a few.
Post McKinsey, Abhinav was at Amazon, where he was Head of Customer Experience, Worldwide and Head of Women’s Clothing business. At Amazon, he drove increased automation using advanced technologies like AI/ML to scale operations and drive growth and profitability. Abhinav lives in the San Francisco – Bay Area and loves to watch and play cricket. He is looking forward to collaborating with you.
Nora Ghaoui examines the limitations of artificial intelligence as it pertains to building a business strategy.
If company strategies risk sounding the same when written by people, what happens when they get written by AI? In this post I examine an AI-generated strategy statement for what it says about the abilities of AI and creating strategies. Three years ago, I asked if large companies all had the same strategy. Perhaps their strategies all sounded the same because managers picked up the same ideas from MBAs and consultants, or because they hired the same copywriters. Last month, a new source of non-differentiating strategy appeared – strategy written by AI.
The AI in question is GPT-3 from OpenAI, which has been getting a lot of attention lately. Here’s a quick introduction to GPT-3: it is a language prediction model that autocompletes text from the input that you give it, like you see when you use Google search. It’s able to complete many different kinds of text, giving it a wider range of application than other models.
Its power comes from its sheer size. It has been trained on a huge amount of text from the internet, and it has 175 billion parameters in which it stores the patterns in that text. Its response to an input is the text that is statistically most likely to come after it. So the more examples it has, the better it can match the input.
Key points covered include:
- Example of strategy written by AI
- How AI and predictive analytics work
- Critical thinking
Read the full article, Can AI Write a Strategy?, on the Veridia website.
Davide Gronchi provides two simple tools that can help collect answers to powerful questions.
Advanced analytics and machine learning are some of the ready-to-use technologies that help discover correlations and drive conclusions out of complex data sets that often describe our business and production processes. This is very helpful to take decisions aiming to prevent something unwanted to happen e.g., set process parameter to X in order to obtain product spec within tolerance.
There are many other opportunities to eliminate “waste” out of business processes that don’t require complex tools and data scientist skills but “just” common sense and good leadership. Solving problems should always start with a clear definition of “what is the problem?” Often we mix up the symptoms with the root causes, by doing so we look for solutions to the symptoms but don’t eliminate the root cause. Guess what? The problem will be back very soon…
Following a structured problem-solving approach is not difficult but requires discipline and asking the right questions, what we call “powerful questions“. These are questions that make people thinking, typically open questions that require an articulated answer, not just a binary yes/no.
Asking powerful questions should be one of the core skills of good leaders: not solving problems themselves but helping their teams to do so. I believe many have forgotten this and risk to lead teams in endless problem solving rounds without sustainable and substantial results.
Included in this article:
- Fishbone diagram
- Pareto chart
Read the full article, The Simple Art of Problem Solving, on the Growing Operations Advisors website.
This article from Karthik Rajagopalan’s company blog introduces a type of machine learning that may provide solutions for some traditional optimization problems such as inventory optimization and supply chain optimization.
The field of artificial intelligence is slowly beginning to permeate our lives. Computers and other machines are being endowed with intelligence through a process called machine learning. One particular type of machine learning that is of interest to us here in Paramis Digital is Reinforcement Learning (RL), which has been quite successful in a limited number of applications like teaching computers how to play video games and teaching robots how to perform certain activities.
Supervised learning uses data with input and labeled output and learns the relationship between them. Unsupervised learning, on the other hand, works on unlabeled data and is generally used for pattern detection. It is not uncommon to find applications that use a combination of the two methods. In reinforcement learning an agent, say a robot, learns to perform activities in an environment to accomplish a specified goal. The agent learns by performing various activities, collecting feedback from the environment in response to the activities and evaluating the feedback. The key part of this learning method is the reward mechanism, a mathematical construct, which rewards the agent for performing activities that will move it towards and eventually accomplish the goal. The agent is also punished for performing activities that detract it from accomplishing its goals.
Read the full article, Reinforcement Learning, on the Paramis Digital website.
Umbrex is pleased to welcome Justin Lechner with Bonsai Growth. Justin Lechner is an ex-Bain consultant based in Brooklyn, NY. After leaving Bain, Justin has split his time between ventures in the enterprise AI space and independent consulting work (innovation and technology projects).
Justin’s particularly passionate about applications of data science within the Consumer, Healthcare, and PE spaces. Justin is excited to collaborate on projects with a data focus
Ben Dattner co wrote this article for Harvard Business Review on the issue of building ethical AI for talent management.
Artificial intelligence has disrupted every area of our lives — from the curated shopping experiences we’ve come to expect from companies like Amazon and Alibaba to the personalized recommendations that channels like YouTube and Netflix use to market their latest content. But, when it comes to the workplace, in many ways, AI is still in its infancy. This is particularly true when we consider the ways it is beginning to change talent management. To use a familiar analogy: AI at work is in the dial-up mode. The 5G WiFi phase has yet to arrive, but we have no doubt that it will.
Areas covered in this article include:
- Training data sets
- Efficient predictions on a candidate
- Bias and creating homogeneity in organizations
Read the full article, Building Ethical AI for Talent Management, on the Harvard Business Review.
Tobias Baer provides clear and concise examples of how Google uses the acquisition of select data to create bias, which leads to the dissemination of inaccurate information.
I’m an avid user of the navigation function of Google Maps. Every time I reach my destination, Google asks me for feedback on the navigation instructions. What could possibly be wrong with that? Well, I bet that the data and any analytics derived from that feedback often – and, vastly! – overestimates users’ satisfaction. Why is that?
The app is a perfect illustration of availability bias. I only am given this opportunity to provide feedback when I reach my destination. Which means that if I reach a river only to find that the ferry supposed to take me and my car to the other riverside has stopped operations an hour ago, or if after a few hours of cycling I find that the path indicated by the app leads straight into a gigantic military infrastructure that is fenced by barbed wires with large red signs threatening any trespasser to be shot (both has actually happened to me), and hence my only option is to abolish my route, exit the navigation, and go back to where I come from, no feedback is collected.
Points covered in this article include:
- The problem with creating algorithms quickly
- The lack of sufficient communication
- The challenge of creating objective, systematic assessment procedures
Read the full article, A Little Example How Google Creates Biases, on LinkedIn.
Karthik Rajagopalan’s company blog explains how machine learning models can facilitate a deeper understanding of the drivers of churn, leading to better solutions that can help customer retention for subscription businesses.
Subscriptions have been around for a very long time. Having come a long way from the hire-for-purchase model introduced by the Singer sewing machine company, the past few decades saw the emergence of memberships in retail, health clubs, and monthly subscriptions to services like telephone and cable television. More recently, the recurring revenue model has been pushed forward by e-commerce and software-as-a-service (SaaS) businesses with the introduction of services for streaming media, connected home, connected car, gaming, etc. According to the Subscription economy index (SEI), the subscription economy has grown by nearly 300% over the last 7 years and is also increasingly correlated with traditional economy, a significant yet not so surprising development.
Topics covered in this article include:
-Customer longterm value
Read the full article, AI augmented retention program is a must for subscription businesses, on the Paramis Digital website.