Many tout AI in ediscovery as the next big thing for the legal tech world – but what can this technology really do for lawyers today?
Can you tell the difference between a picture of a cute brown chihuahua and a tasty blueberry muffin? Most humans can do this easily. (But it’s harder than you might think – take a look here.) Can a computer tell the difference? Only if you correctly train it to do so – and once it’s trained, the answers can be objectively accurate. This is the simple premise behind “artificial intelligence,” or AI, which is the science of teaching computers how to “learn, reason, perceive, infer, communicate, and make decisions like humans do.”
Every Day AI
Spell-check is a decades-old function you probably take advantage of in your favorite word processing software. How does the computer know you spelled a word wrong? Because it’s been “trained” using a dictionary. But in traditional spell-checkers, this dictionary is all the computer has to go by – you can manually add new words to it, but it won’t learn anything new on its own.
Now, word processors use AI to give spell-check tools wider capabilities. Some platforms will self-train new spellings based on the number of times that you use a unique word and how often you manually correct it. Word processors also make suggestions for grammar and syntax, using AI to understand the intricacies of written language that can’t be conveyed in a simple dictionary of words and spelling.
How does Netflix know what movies or shows to suggest you watch next? It’s based on the shows you’ve already binged, and compared with hundreds of millions of other viewers that watched similar shows and movies that the Netflix mothership computers “learned” from. Dozens of other technologies that you interact with everyday use the same methodology, like Amazon, Siri, spam recognition servers, and even the Photos app on your phone. (Can it recognize a chihuahua vs. a blueberry muffin?)
When it comes to the legal profession, AI can be similarly used for document filing. If a document management system watches how you regularly file documents or emails from a particular client, it can train itself to suggest that same filing protocol on other communications with that client. Or computers can be trained to compose contracts with similar clauses and sections based on past approved contracts. This could be even more acceptable and accurate than the “manual” process of a lawyer pulling up an old contract and attempting to change all the old names and dates, which often means something gets overlooked.
AI is NOT a Replacement for Human Expertise – It’s an Enhancement
In order to successfully embrace the practical applications of AI in ediscovery, we have to shift away from a mindset of “robots replacing humans,” which is a storyline that only exists in science fiction movies. AI is a tool that helps you get your job done, like a pencil, a stapler, a printer or fax machine (if you still have one).
As legal journalist Bob Ambrogi stated simply, “AI is a tool. A tool that we control.
It’s a tool that can make us more effective and efficient at what we do.
It’s a tool that can help us deliver our services more quickly and at lower cost.
It’s a tool that can help us serve more clients and serve them better.
It’s a tool that can enhance the delivery of legal services and that can help to close the yawning justice gap.
It’s a tool for good, not for evil.”
At the Nextpoint On Point User Conference in September 2022, we invited legal technology thought leaders to participate in a panel titled “The Future of Legal Technology and Artificial Intelligence,” and the discussion echoed Ambrogi’s thoughts.
Joey Gartner, Director of the ABA’s Center for Innovation, stated, “Technology is a tool. It will not solve all of your problems, but it will solve some of your problems.”
Dan Linna, Director of Law and Technology Initiatives at Northwestern Pritzker School of Law and McCormick School of Engineering, added, “These tools can help us not just be more efficient, but improve the quality of our work product and give us better outcomes.”
Every legal professional is familiar with the hours spent in a law library researching case law, tracking down the best opinions and articles to support or refute a legal stance. What if a computer could be trained to understand a legal query and produce a set of research results that could achieve your goal? Of course, you wouldn’t trust those initial results until you verified them and then possibly further trained the computer to provide more accurate results. The focus is not specifically on “automating” the task of legal research, but “augmenting” the decisions that are ultimately made by humans.
Objective vs. Subjective Decisions
Perhaps the friction with AI in the legal profession comes from how the decisions are made. Answering whether or not a picture shows a chihuahua or a blueberry muffin is an objective question – it’s either a dog or a baked good, and every sensible human will answer the question the same way.
When it comes to making certain decisions in the legal field, like whether a document is relevant or not to a litigation matter, many lawyers argue that kind of decision can only be made subjectively based on their [human] years of experience and professional knowledge. Those lawyers can’t fathom that such a decision could be made by a computer… and they would be 100% correct, unless that lawyer “trained” a computer by highlighting relevant or “hot” documents, which would allow the computer to recognize similar relevant documents. And the computer can make those decisions without taking a lunch break or involving emotional attachments or sentiment.
But no litigator worth their salt would allow documents identified as “relevant” by a computer to be produced without looking them over first. In other words, the computer hasn’t replaced the litigator’s stance and experience, it has simply assisted by providing a starting point, or at least a jumping off point to identify a corpus of data and documents and files. At the very least, the computer could prioritize a list of documents that need to be reviewed that could save a lot of wasted time and effort. Again, it’s not “artificial” intelligence, it’s “augmented” intelligence.
Sonali Ray, the Director of Legal Strategy at Nextpoint, participated in the On Point AI panel and encouraged litigators to “work in conjunction with the technology – look at it as an assist, not an ‘easy button.’ The human eye is not perfect. Even though there are elements of the machine that aren’t perfect, neither is your technique.”
Dan Linna echoed Sonali’s comments by stating, “We still hold onto this mythology that humans (especially lawyers) are perfect in the work they’re doing. But studies show that humans are not perfect. There are huge opportunities to use computation to augment the work that humans are doing.”
Bruce Fein, the Legal Director and co-founder of Dagger Analytics, provided a helpful definition of how AI can help in the realm of ediscovery and litigation: “In the legal context, AI is about getting a computer to make a decision, like relevant or not relevant. These tools are intelligently moving toward the best algorithm for making that decision. Rather than you creating that algorithm, the computer is figuring it out for you.”
From Artificial Chihuahuas to AI in Ediscovery
In concluding the On Point AI panel, Dan Linna shared a vision for where AI could enhance the legal profession: “All too often we think about using AI to do things that we’ve always done, just a little bit better, a little bit faster, a little bit cheaper. But we need to change our mindset to think about how we can use AI to transform the things that we do. We need to start thinking more about some of the other AI principles that are being discussed in other areas that AI is being used – fairness, transparency, accountability.”
In order for the legal profession to successfully integrate AI tools into ediscovery projects, it’s important to understand that the technology is simply a tool that can help us better accomplish certain tasks, as well as transform and reshape legacy, manual approaches to litigation. And in the process, AI can help us avoid any confusion between chihuahuas v. blueberry muffins.
See more »
Refine your interests »
Back to Top
Explore 2022 Readers’ Choice Awards
Copyright © var today = new Date(); var yyyy = today.getFullYear();document.write(yyyy + ” “); JD Supra, LLC

source

bauna/iStock 
Microsoft-backed OpenAI's AI chatbot ChatGPT has been making headlines ever since it was released to the public on November 30. It can break down complex scientific concepts, compose poems, write stories, code, and create malware…the list is endless. OpenAI has also released a paid version of the chatbot. Known as 'ChatGPT Professional', it is available at $42 per month.
ChatGPT3 has even raised alarms within Google – according to a recent report by The New York Times, 'Google now intends to unveil more than 20 new products and demonstrate a version of its search engine with chatbot features this year'.
Students have also been using the chatbot to complete assignments. It turns out it can clear examinations, too, with flying colors. Christian Terwiesch, a professor at the Wharton School School of Business, University of Pennsylvania, tested the performance of ChatGPT in an MBA exam. He questioned the chatbot on Operations Management, a core MBA subject. 
The paper, titled 'Would Chat GPT3 Get a Wharton MBA? A Prediction Based on Its Performance in the Operations Management Course', delved into the implications of ChatGPT's 'academic performance'.
Christian Terwiesch  
Based on the test, Terwiesch found that the answers were correct, and the chatbot provided excellent explanations. However, ChatGPT also made mistakes in relatively simple calculations at the sixth-grade math level, which can be massive in magnitude. 
Terwiesch also noted that the current version of ChatGPT cannot handle more advanced process analysis questions, even if they are based on fairly standard templates. 
Terwiesch concluded by stating in his study that ChatGPT has "remarkable skills in handling problems as used extensively in the training and testing of our MBA students. Combining the results of the questions, I would grade this performance as a B to B-". 
The professor also added a reference point to put the chatbot's performance into perspective: "Until Wharton allowed students more flexibility in which courses they take, this Operations Management course was a required course that every student had to take. However, we did allow students to waive this course if they could demonstrate content mastery on a waiver exam. The performance of Chat GPT3 reported above would have been sufficient to pass the waiver exam, though by a very small margin."
Christian Terwiesch 
Terwiesch didn't stop there. The professor wanted to find out if Chat GPT3 could develop a question paper. "By now, I have written thousands of questions, and, at times, I feel I have exhausted my imagination for new problems. Can I turn to Chat GPT3 to come up with new exam questions?" he asks in the study.
It was found that the questions were plausible and humorous. Terwiesch mentioned that the questions were good enough to be taken advantage of in upcoming question papers.
Professor Terwiesch compared the effect electronic calculators had on the corporate world to the impact ChatGPT could have on academia. "Prior to the introduction of calculators and other computing devices, many firms employed hundreds of employees whose task it was to manually perform mathematical operations such as multiplications or matrix inversions. Obviously, such tasks are now automated, and the value of the associated skills has dramatically decreased. In the same way, any automation of the skills taught in our MBA programs could potentially reduce the value of an MBA education," he said in the study.
Andrew Karolyi, dean of Cornell University’s SC Johnson College of Business, resonated with the same. He told the Financial Times: "One thing we all know for sure is that ChatGPT is not going away. If anything, these AI techniques will continue to get better and better. Faculty and university administrators need to invest to educate themselves."
Study Abstract:
OpenAI’s Chat GPT3 has shown a remarkable ability to automate some of the skills of highly compensated knowledge workers in general and specifically the knowledge workers in the jobs held by MBA graduates including analysts, managers, and consultants. Chat GPT3 has demonstrated the capability of performing professional tasks such as writing software code and preparing legal documents. The purpose of this paper is to document how Chat GPT3 performed on the final exam of a typical MBA core course, Operations Management. Exam questions were uploaded as used in a final exam setting and then graded. The “academic performance” of Chat GPT3 can be summarized as follows. First, it does an amazing job at basic operations management and process analysis questions including those that are based on case studies. Not only are the answers correct, but the explanations are excellent. Second, Chat GPT3 at times makes surprising mistakes in relatively simple calculations at the level of 6th grade Math. These mistakes can be massive in magnitude. Third, the present version of Chat GPT is not capable of handling more advanced process analysis questions, even when they are based on fairly standard templates. This includes process flows with multiple products and problems with stochastic effects such as demand variability. Finally, ChatGPT3 is remarkably good at modifying its answers in response to human hints. In other words, in the instances where it initially failed to match the problem with the right solution method, Chat GPT3 was able to correct itself after receiving an appropriate hint from a human expert. Considering this performance, Chat GPT3 would have received a B to B- grade on the exam. This has important implications for business school education, including the need for exam policies, curriculum design focusing on collaboration between humans and AI, opportunities to simulate real-world decision-making processes, the need to teach creative problem-solving, improved teaching productivity, and more.
Known as ART, the amphibious robot could help with monitoring challenging terrestrial-aquatic ecosystems.

source

Copyright © 2023 InvestorPlace Media, LLC. All rights reserved. 1125 N. Charles St, Baltimore, MD 21201.
With the stock market constantly fluctuating, it can be difficult to know which stocks to sell and which ones to hold onto. But you can avoid these three for now
Source: Shutterstock
With tech stocks continuing to rise, it is becoming increasingly difficult to decide which companies are worth buying, and which are simply stocks to sell. This article will give readers an overview of the best tech stocks to sell to maximize their returns.
The U.S., European, and Chinese stock markets have experienced positive gains since the start of the year. However, despite this recent bullishness, UBS Global Wealth Management cautioned against being over-confident in the sustainability of this run. Factors like high inflation and other market conditions could still be unfavorable for stocks in the early months of 2023.
Mark Haefele, Chief Investment Officer at UBS Global Wealth Management, voiced his concern over the possibility of a ‘head fake’ rally, and that economic data may not achieve the market’s expectations in a recent note to clients. He cautioned that it’s still too soon to infer that inflation is no longer a concern. Additionally, he highlighted the possibility of core inflation being higher than anticipated, along with other potential risks facing the markets.
Investors could not be happier with the positive start to this year. However, they should also remain watchful. Although the market is looking up, economic data are still uncertain. Thus, it’s far from guaranteed that this impressive progress we’ve seen will remain for the rest of 2023.
Accordingly, for those looking to trim equity exposure, here are three stocks to sell.
DocuSign (NASDAQ:DOCU) is a company providing digital signature solutions to a broad base of large and small corporate clients. This business model has made the company one of the most sought-after tech stocks during the pandemic. Indeed, as businesses of all sizes adjusted their operations as a result of the pandemic, many leaned on digital solutions like electronic signatures and the document management tools that DocuSign offers.
DocuSign’s yearly revenue has seen tremendous growth in the last three years. In 2022, the company reported $2.1 billion in revenue, a 45% increase on a year-over-year basis. Impressively, 2021’s $1.453 billion in revenue was also roughly 50% higher over 2020, meaning this is a compounder with some serious clout. That said, revenue growth has slowed of late, with the company reporting top-line growth of 24.5% for the 12 months ended Oct. 31, 2022.
Growth has slowed further, to just 18%, as pr the company’s recently-released Q3 and fiscal 2023 financial results. Subscription income came in at $624.1 million, an increase of 18% compared to the year prior. Professional services and other revenue registered a boost of 27%, amounting to $21.4 million compared to the same period last year. However, the numbers signify a decrease sequentially, and reflect a general decline in growth for this previous high-flyer.
In addition, the dip in the residential real estate market is a cause for worry. When he published his piece on tech stocks to sell in December, InvestorPlace contributor Larry Ramer made an astute evaluation. That is, that the housing market was among the driving forces behind this organization’s success. The data proves Ramer is right.
Unfortunately, the US housing market saw another decline in December, extending the slump to four consecutive months in 2022. This marked a difficult year for the industry, which experienced its first annual decrease in housing starts since 2009.
Many people, including Larry, used the software when purchasing a house. However, the market downturn has intensified downward pressure on DocuSign, which is why it is on this list of tech stocks to sell.
Opendoor Technologies (NASDAQ:OPEN) is bringing about a revolution in the home-buying process with its disruptive technology. It aims to provide an automated solution for a smoother, quicker, and more convenient buying experience. Accordingly, it’s no surprise to see the influx of investors to this stock, when it made its debut in 2020.
In 2020, when Opendoor made its stock market debut, investors swarmed to the investment opportunity. This was at the pandemic’s peak, when investors were flush with cash and looking for a place to grow it. As a result, the stock did very well during its initial few weeks, surging in value as speculators entered the market.
However, Opendoor’s stock price has hit a rough patch over the past year. This is primarily due to increasingly bearish market sentiment. OPEN stock has lows two-thirds of its value over the past year, with expectations building that more in the way of declines could be on the horizon.
That’s largely due to the widespread aforementioned decline in the real estate market. Higher interest rates have killed this market, with home starts seeing one of the worst declines on record. Redfin anticipates that there will be a 16% decline in the number of existing home sales from 2022 to 2023, resulting in 4.3 million total sales. According to the company’s report, buyers are hesitant to make purchases due to affordability issues such as inflation, higher mortgage rates, and pricey homes, along with the possibility of an economic recession. Morgan Stanley (NYSE:MS) experts are also anticipating a fall in the housing sector by 2023, which could be damaging to those who bought their homes the previous year in 2022.
Undoubtedly, Opendoor’s business model is disruptive. But market trends are going against the stock, making this a top stock to sell in my books right now.
Ah, how time flies! It seems like yesterday we were all discussing Silvergate Capital (NYSE:SI), a Californian bank that mainly specializes in cryptocurrency transactions. However, after the epic downturn in the crypto markets and the spectacular collapse of FTX, Silvergate Capital is on the ropes.
On Jan. 17, Silvergate Capital revealed its fiscal Q4 earnings, recording a net loss of $1.0 billion or ($33.16 per share). Average digital asset deposits declined to $7.3 billion from the prior quarter’s $12.0 billion. Following these results, investors have clearly priced in worries about a run on the bank, which could lead to a collapse in Silvergate Capital in short order. Fortunately, this hasn’t occurred yet, due in part to the company’s reported total deposits of $3.8 billion at the end of the quarter.
That said, during the quarter, management reported $5.2 billion in sales of debt and securities at a disadvantageous expense of $718 million, to ensure sufficient liquidity. The firm reported a massive loss, and the company’s stock price reflected this reality as well.
Those who think that this lower stock price provides a great entry point should be warned. The selling pressure with SI stock may be far from over. Many investors didn’t think the company will be able to make it out of this crypto winter. And while Silvergate Capital may continue to sustain itself temporarily on trading fees from its exchange-traded products, it’s unclear how much investor demand will remain for its shares, should another contagion event take place.
On the publication date, Faizan Farooque did not hold (either directly or indirectly) any positions in the securities mentioned in this article. The opinions expressed in this article are those of the writer, subject to the InvestorPlace.com Publishing Guidelines.
Faizan Farooque is a contributing author for InvestorPlace.com and numerous other financial sites. Faizan has several years of experience in analyzing the stock market and was a former data journalist at S&P Global Market Intelligence. His passion is to help the average investor make more informed decisions regarding their portfolio.
Financial, Fintech, Real Estate, Technology, Software
Growth Stocks

Article printed from InvestorPlace Media, https://investorplace.com/2023/01/3-tech-stocks-to-sell-in-january-before-they-get-torpedod-docu-open-si/.
©2023 InvestorPlace Media, LLC
Stocks to Sell
Hot Stocks
Market Analysis
Today's Market
Today's Market
Stocks to Sell
Stocks to Buy
Today's Market
Financial Market Data powered by FinancialContent Services, Inc. All rights reserved. Nasdaq quotes delayed at least 15 minutes, all others at least 20 minutes. Copyright © 2023 InvestorPlace Media, LLC. All rights reserved. 1125 N. Charles St, Baltimore, MD 21201.
Not Yet a Premium Subscriber?

source

source

Video Management Software Market
Permanent link to this press release:
Delete press release Edit press release
All 5 Releases

source

BrandPosts are written and edited by members of our sponsor community. BrandPosts create an opportunity for an individual sponsor to provide insight and commentary from their point-of-view directly to our audience. The editorial team does not participate in the writing or editing of BrandPosts.
It is said that Switzerland is a country where residents book appointments to do laundry in their own homes.
In other words, they leave little for chance.
But when the unexpected intruded on the mountain nation, the Canton – or state – of Zurich had to scramble to deal with the aftershock of the COVID pandemic.
Almost overnight, the number of applications for short-time work rose to more than 10,000 – up from 10 in the months before.
In a three month stretch, 30,000 payments had to be processed – a task that previously took 25 minutes per imbursement.
In normal times, an applicant would have had to download an Excel file, fill out certain parts by hand, and send it out by mail. The form would then be scanned and uploaded to a 30-year-old document management system with a complex interface.
Even then, much of the data was inaccurate or incomplete.
The result-oriented Swiss were not about to revert to these flawed methods when so much was at stake. The flurry of applications would have required the canton to hire an additional 70 full-time employees – a near-impossible undertaking when so much of the country was locked down.
But what if humans weren’t brought on? 
What if the Canton of Zurich turned to robotics instead?
Human ingenuity and automated efficiency
With a population of more than 1.5 million, the Canton of Zurich is considered the heart of Switzerland’s economy. And as the global center of banking and finance, the city of Zurich is the canton’s de facto capital.
As in the rest of Switzerland, citizens appear to be obsessed with getting things right.
Determined not to make any mistakes as it developed the short-time work application and payment solution, the canton began working with enterprise resource planning (ERP) software leader SAP. SAP’s Business Technology Platform (BTOP) would handle basic and backend functionalities, while SAP’s Intelligent Robotic Process Automation (iRPA) provided the software robots and digital workers.
Working under a tight deadline, SAP Services and Support mobilized a versatile, global team specializing in flexible problem-solving. Once targets were agreed upon each morning, the iRPA developers began coding various parts of the application.
The iRPA was linked to two older legacy systems, eliminating the need to create expensive and time-consuming interfaces.
Every evening, business experts conducted application tests, delivering feedback to the developers and project team.
Meticulous timing
The platform faced close scrutiny when it was deployed in April 2020. Within two weeks, the processing time per application had decreased from 25 minutes to 30 seconds. How’s that for Swiss precision?
Two bots took on the primary responsibilities for automating the applications and payments, as well as ensuring that most data was digitally authenticated according to local business rules. 
Payments were received in less than two weeks.
In total, the canton was able to rely on an automation level of 85%, with just 15% of the documents having to be revalidated by a live human being.
No additional staff members needed to be recruited and employed during a period when most Swiss citizens were working from home.
Without the solution, “we would never have managed to process and pay the enormous number of short-time working compensations,” observed Christian Truog, CFO of the Canton of Zurich’s labor authority.
Most importantly, the canton now understands the fine points of developing a modern iRPA application – knowledge that can be called upon to meet the next challenge.
As a result, the office of information technology at the Canton of Zurich’s department of finance received a 2022 SAP Innovation Award, based on the pragmatic project approach, swift implementation, business benefits and positive feedback on the new platform. You can get the details on what they accomplished to earn this coveted award in their pitch deck.
Despite the pressures and the quick turnover, the solution managed to embody everything about the Swiss character – a culture where, as the adage goes, spontaneity is considered wonderful, so long as it’s planned.
Sponsored Links
News
Reviews
Buyer’s Guides
Blogs/Opinion
Podcasts
Awards programs
View the archive

source

source

source

WorldView and Homecare Homebase Deepen Their Partnership to …  PR Web
source

Generative AI models for businesses threaten to upend the world of content creation, with substantial impacts on marketing, software, design, entertainment, and interpersonal communications. These models are able to produce text and images: blog posts, program code, poetry, and artwork. The software uses complex machine learning models to predict the next word based on previous word sequences, or the next image based on words describing previous images. Companies need to understand how these tools work, and how they can add value.
Large language and image AI models, sometimes called generative AI or foundation models, have created a new set of opportunities for businesses and professionals that perform content creation. Some of these opportunities include:

How adept is this technology at mimicking human efforts at creative work? Well, for an example, the italicized text above was written by GPT-3, a “large language model” (LLM) created by OpenAI, in response to the first sentence, which we wrote. GPT-3’s text reflects the strengths and weaknesses of most AI-generated content. First, it is sensitive to the prompts fed into it; we tried several alternative prompts before settling on that sentence. Second, the system writes reasonably well; there are no grammatical mistakes, and the word choice is appropriate. Third, it would benefit from editing; we would not normally begin an article like this one with a numbered list, for example. Finally, it came up with ideas that we didn’t think of. The last point about personalized content, for example, is not one we would have considered.
Overall, it provides a good illustration of the potential value of these AI models for businesses. They threaten to upend the world of content creation, with substantial impacts on marketing, software, design, entertainment, and interpersonal communications. This is not the “artificial general intelligence” that humans have long dreamed of and feared, but it may look that way to casual observers.
Generative AI can already do a lot. It’s able to produce text and images, spanning blog posts, program code, poetry, and artwork (and even winning competitions, controversially). The software uses complex machine learning models to predict the next word based on previous word sequences, or the next image based on words describing previous images. LLMs began at Google Brain in 2017, where they were initially used for translation of words while preserving context. Since then, large language and text-to-image models have proliferated at leading tech firms including Google (BERT and LaMDA), Facebook (OPT-175B, BlenderBot), and OpenAI, a nonprofit in which Microsoft is the dominant investor (GPT-3 for text, DALL-E2 for images, and Whisper for speech). Online communities such as Midjourney (which helped win the art competition), and open-source providers like HuggingFace, have also created generative models.
These models have largely been confined to major tech companies because training them requires massive amounts of data and computing power. GPT-3, for example, was initially trained on 45 terabytes of data and employs 175 billion parameters or coefficients to make its predictions; a single training run for GPT-3 cost $12 million. Wu Dao 2.0, a Chinese model, has 1.75 trillion parameters. Most companies don’t have the data center capabilities or cloud computing budgets to train their own models of this type from scratch.
But once a generative model is trained, it can be “fine-tuned” for a particular content domain with much less data. This has led to specialized models of BERT — for biomedical content (BioBERT), legal content (Legal-BERT), and French text (CamemBERT) — and GPT-3 for a wide variety of specific purposes. NVIDIA’s BioNeMo is a framework for training, building and deploying large language models at supercomputing scale for generative chemistry, proteomics, and DNA/RNA.OpenAI has found that as few as 100 specific examples of domain-specific data can substantially improve the accuracy and relevance of GPT-3’s outputs.
To use generative AI effectively, you still need human involvement at both the beginning and the end of the process.
To start with, a human must enter a prompt into a generative model in order to have it create content. Generally speaking, creative prompts yield creative outputs. “Prompt engineer” is likely to become an established profession, at least until the next generation of even smarter AI emerges. The field has already led to an 82-page book of DALL-E 2 image prompts, and a prompt marketplace in which for a small fee one can buy other users’ prompts. Most users of these systems will need to try several different prompts before achieving the desired outcome.
Then, once a model generates content, it will need to be evaluated and edited carefully by a human. Alternative prompt outputs may be combined into a single document. Image generation may require substantial manipulation. Jason Allen, who won the Colorado “digitally manipulated photography” contest with help from Midjourney, told a reporter that he spent more than 80 hours making more than 900 versions of the art, and fine-tuned his prompts over and over. He then improved the outcome with Adobe Photoshop, increased the image quality and sharpness with another AI tool, and printed three pieces on canvas.
Generative AI models are incredibly diverse. They can take in such content as images, longer text formats, emails, social media content, voice recordings, program code, and structured data. They can output new content, translations, answers to questions, sentiment analysis, summaries, and even videos. These universal content machines have many potential applications in business, several of which we describe below.
These generative models are potentially valuable across a number of business functions, but marketing applications are perhaps the most common. Jasper, for example, a marketing-focused version of GPT-3, can produce blogs, social media posts, web copy, sales emails, ads, and other types of customer-facing content. It maintains that it frequently tests its outputs with A/B testing and that its content is optimized for search engine placement. Jasper also fine tunes GPT-3 models with their customers’ best outputs, which Jasper’s executives say has led to substantial improvements. Most of Jasper’s customers are individuals and small businesses, but some groups within larger companies also make use of its capabilities. At the cloud computing company VMWare, for example, writers use Jasper as they generate original content for marketing, from email to product campaigns to social media copy. Rosa Lear, director of product-led growth, said that Jasper helped the company ramp up our content strategy, and the writers now have time to do better research, ideation, and strategy.
Kris Ruby, the owner of public relations and social media agency Ruby Media Group, is now using both text and image generation from generative models. She says that they are effective at maximizing search engine optimization (SEO), and in PR, for personalized pitches to writers. These new tools, she believes, open up a new frontier in copyright challenges, and she helps to create AI policies for her clients. When she uses the tools, she says, “The AI is 10%, I am 90%” because there is so much prompting, editing, and iteration involved. She feels that these tools make one’s writing better and more complete for search engine discovery, and that image generation tools may replace the market for stock photos and lead to a renaissance of creative work.
DALL-E 2 and other image generation tools are already being used for advertising. Heinz, for example, used an image of a ketchup bottle with a label similar to Heinz’s to argue that “This is what ‘ketchup’ looks like to AI.” Of course, it meant only that the model was trained on a relatively large number of Heinz ketchup bottle photos. Nestle used an AI-enhanced version of a Vermeer painting to help sell one of its yogurt brands. Stitch Fix, the clothing company that already uses AI to recommend specific clothing to customers, is experimenting with DALL-E 2 to create visualizations of clothing based on requested customer preferences for color, fabric, and style. Mattel is using the technology to generate images for toy design and marketing.
GPT-3 in particular has also proven to be an effective, if not perfect, generator of computer program code. Given a description of a “snippet” or small program function, GPT-3’s Codex program — specifically trained for code generation — can produce code in a variety of different languages. Microsoft’s Github also has a version of GPT-3 for code generation called CoPilot. The newest versions of Codex can now identify bugs and fix mistakes in its own code — and even explain what the code does — at least some of the time. The expressed goal of Microsoft is not to eliminate human programmers, but to make tools like Codex or CoPilot “pair programmers” with humans to improve their speed and effectiveness.
The consensus on LLM-based code generation is that it works well for such snippets, although the integration of them into a larger program and the integration of the program into a particular technical environment still require human programming capabilities. Deloitte has experimented extensively with Codex over the past several months, and has found it to increase productivity for experienced developers and to create some programming capabilities for those with no experience.
In a six-week pilot at Deloitte with 55 developers for 6 weeks, a majority of users rated the resulting code’s accuracy at 65% or better, with a majority of the code coming from Codex.  Overall, the Deloitte experiment found a 20% improvement in code development speed for relevant projects. Deloitte has also used Codex to translate code from one language to another. The firm’s conclusion was that it would still need professional developers for the foreseeable future, but the increased productivity might necessitate fewer of them. As with other types of generative AI tools, they found the better the prompt, the better the output code.
LLMs are increasingly being used at the core of conversational AI or chatbots. They potentially offer greater levels of understanding of conversation and context awareness than current conversational technologies. Facebook’s BlenderBot, for example,  which was designed for dialogue, can carry on long conversations with humans while maintaining context. Google’s BERT is used to understand search queries, and is also a component of the company’s DialogFlow chatbot engine. Google’s LaMBA, another LLM, was also designed for dialog, and conversations with it convinced one of the company’s engineers that it was a sentient being— an impressive feat, give that it’s simply predicting words used in conversation based on past conversations.
None of these LLMs is a perfect conversationalist. They are trained on past human content and have a tendency to replicate any racist, sexist, or biased language to which they were exposed in training. Although the companies that created these systems are working on filtering out hate speech, they have not yet been fully successful.
One emerging application of LLMs is to employ them as a means of managing text-based (or potentially image or video-based) knowledge within an organization. The labor intensiveness involved in creating structured knowledge bases has made large-scale knowledge management difficult for many large companies. However, some research has suggested that LLMs can be effective at managing an organization’s knowledge when model training is fine-tuned on a specific body of text-based knowledge within the organization. The knowledge within an LLM could be accessed by questions issued as prompts.
Some companies are exploring the idea of LLM-based knowledge management in conjunction with the leading providers of commercial LLMs. Morgan Stanley, for example, is working with OpenAI’s GPT-3 to fine-tune training on wealth management content, so that financial advisors can both search for existing knowledge within the firm and create tailored content for clients easily. It seems likely that users of such systems will need training or assistance in creating effective prompts, and that the knowledge outputs of the LLMs might still need editing or review before being applied. Assuming that such issues are addressed, however, LLMs could rekindle the field of knowledge management and allow it to scale much more effectively.
We have already seen that these generative AI systems lead rapidly to a number of legal and ethical issues. “Deepfakes,” or images and videos that are created by AI and purport to be realistic but are not, have already arisen in media, entertainment, and politics. Heretofore, however, the creation of deepfakes required a considerable amount of computing skill. Now, however, almost anyone will be able to create them. OpenAI has attempted to control fake images by “watermarking” each DALL-E 2 image with a distinctive symbol. More controls are likely to be required in the future, however — particularly as generative video creation becomes mainstream.
Generative AI also raises numerous questions about what constitutes original and proprietary content. Since the created text and images are not exactly like any previous content, the providers of these systems argue that they belong to their prompt creators. But they are clearly derivative of the previous text and images used to train the models. Needless to say, these technologies will provide substantial work for intellectual property attorneys in the coming years.
From these few examples of business applications, it should be clear that we are now only scratching the surface of what generative AI can do for organizations and the people within them. It may soon be standard practice, for example, for such systems to craft most or all of our written or image-based content — to provide first drafts of emails, letters, articles, computer programs, reports, blog posts, presentations, videos, and so forth. No doubt that the development of such capabilities would have dramatic and unforeseen implications for content ownership and intellectual property protection, but they are also likely to revolutionize knowledge and creative work. Assuming that these AI models continue to progress as they have in the short time they have existed, we can hardly imagine all of the opportunities and implications that they may engender.

source