Cloud Hosting

This week we met with Workday at the companys annual Innovation Summit and I walked away very impressed. Not only is Workday clear-eyed and definitive about its AI product strategy, the company is entering one of its strongest product cycles in years. I have never seen so many Workday features reach maturity and its clear to me that the platform is hitting on all cylinders.

Let me start with an overview: the ERP market is big, important, and changing. Every company needs a financial and human capital system, and these platforms are being asked to do hundreds of things at once. We expect them to be easy to use, fast, and instantly configurable for our company. But we also want them to be easy to extend, open to integration with many other systems, and built on a modern architecture. How can Workday, a company founded 18 years ago, stay ahead in all these areas?

Its actually pretty simple. Workday is not an ERP or software applications company: its a technology company that builds platforms for business solutions. In other words, Workday thinks architecture first, applications second, and this was reinforced again and again as we went through Workdays offerings. Let me give you a few insights on what we learned, and I encourage you to contact us or read more from Workday on many of the things below.

First, Workday is quite clear that AI and Machine Learning will, over time, reinvent what business systems do. The traditional ERP world was a set of core business applications which include Financials, Human Capital (HCM), Supply Chain, Manufacturing, and later Marketing, Customer Analysis, and others. Almost every vendor who starts in one of these areas tries to move into adjacencies, primarily with the goal of selling more software to existing customers.

Today, while companies want to consolidate these applications (a big opportunity for Workday), the bigger goal is reinventing how these applications work together. As Workday describes it, their goal is to help businesses improve planning, execution, and analysis. When its hard to hire, like it will likely continue to be for years, we want the HCM system to help us find contractors, look at alternative work arrangements, and arrange financial and billing solutions to outsource work or tasks, and also find and develop internal candidates. So the red lines between these applications is blurring, and Workday understands this well.

In a sense this is the core of our new Systemic HR Operating Model. We want these various HCM systems, for example, to look at all four of these elements and help us manage them together. Workdays new HCM demo actually showed some of this in action.

Beyond ERP To AI And ML At The Core

But the platform market is even moving faster. Not only do companies want a suite of apps that work together (Workday, Oracle, SAP, and others do this), they want AI and machine learning to operate across the company. And this will change what ERP systems do. Workday listed more than 50 different machine learning experiences the company is already delivering, and the take the form of recommendations or forms pre-filled out or workflows pre-designed that dont look like magic, they just look like intelligent systems that help you run your company better. And this is where Workday is focused.

The new Workforce Management system (labor optimization), for example, can predict hiring and staffing needs based on month, weather, and other external inputs. It can then schedule workers based on their availability, skills, and wages. And it can automatically create a workforce schedule, decide when contract labor is needed, and then automatically create hiring portals and candidate experiences to find people. This is really AI-enabled ERP not a fancy demo of Generative AI to make emails easier to write.

Workday HCM Continues To Mature

The Workday HCM suite is in the strongest shape Ive seen in years. The Workday Skills Cloud is maturing into a skills intelligence platform and it now has features that make it almost essential for a Workday customer. It can import data from any vertical or specialized skills database, it gives companies multiply ways to infer or assess skills, and it gives you dozens of ways to report on skills gaps, predict skills deficiencies, and create upskilling pathways for each employee or workforce group. Ive watched this technology grow over the years and never before have I seen it so well put together and positioned to do what companies want.

This is not to say, by the way, that companies still need specialized skills systems for recruiting (Eightfold, Beamery, Phenom, Seekout, Paradox, iCims, others), mobility (Gloat, Fuel50), learning (Cornerstone, Docebo, Degreed), pay equity (Syndio, Trusaic, Salary.com), and many more. In some sense every HR tech platform now has a skills engine under the covers (remember, a skill is a series of words that describes attributes of a person) and these systems leverage these data elements for very unique purposes. Skills Cloud, in its more mature position in the market, is intended to be a consolidation point to bring the taxonomy into one place. (And its the skills engine that the Workday HCM tools rely upon.)

I know, by the way, that all Workday customers have a multitude of other HCM systems. Given the innovation cycle taking place (vendors are getting on the AI bandwagon in very creative ways), this is going to continue. But Workdays role as the core remains strong, particularly because of my next point.

Workday Is Now Truly Open

I was also impressed with Workdays progress with Extend and Orchestrate, the external APIs and development tools that enable customers and partners to build add-on applications. Workday as a company is not planning on building a lot of vertical solutions, rather they are now pushing partners (Accenture, PwC, and clients) to contribute to the app ecosystem. This creates a force multiplier effect where third parties can make money by building a dev team around Workday. (This, by the way, is why Microsoft is so ubiquitous: their reseller and partner network is massive.)

In addition to these programming interfaces, Workday has made a serious commitment to Microsoft Teams (Workday Everywhere). You can now view Workday cards within Teams and click on deep links within Teams that take you right to Workday transactions. While the company is still committed to continuous improvements in its user interface, I think Workday now understands that users will never spend all day figuring out how Workday works. I believe this trend will continue, and I encouraged Workday to consider Chat-GPT as the next major interface to build. (They were non-commital).

Vertical Applications

I asked the management team what do you think about Oracles decision to buy Cerner, one of the leaders in clinical patient management? Do you think this threatens your vertical strategy? Aneel Bhusri jumped up to argue we would never buy an old legacy company like that it would never integrate into our architecture. This matters because Workdays integrated architecture lets the company deliver AI at scale. In other words, Workday intends to be the pure-play architectural leader, and let the vertical applications come over time.

Today Workday focuses on the education market and has several vertical solutions in financial services, insurance, and healthcare (many built by partners). I dont think the company is going to follow the SAP or Oracle strategy to build deep vertical apps. And this strategy, that of staying pure to the core architecture, may play out well in the longrun. So for those of you who want to build addons, Workday is opening up faster than ever.

What About AI In The Core?

Now lets talk about AI, the most important technology innovation of our time. Sayan Chakraborty, the new co-president and a recognized academic expert on AI, has a very strong position. He believes that Workdays 60 million users (many of which have opted in to be used for anonymous neural network analysis), give the company a massive AI-enabled platform already. So the companys strategy is to double down on declarative AI (machine learning) and then look at Generative AI as a new research effort.

In many ways Workday as been doing AI since they acquired Identified in 2014, and many AI algorithms are built into the Skills Cloud, sourcing and recruiting tools, and myriad of tools for analytics, adaptive planning, and learning. Most of the product managers have AI-related features on their plates, and David Somers, who runs the HCM suite, told us there are hundreds of ideas for new AI features floating around. So in many ways Workday has been an AI platform for years: theyre just now starting to market it.

That said, Workdays real data assets are not that big. Assume that 30 million Workday users have opted in to Workdays AI platform. And lets assume that the Skills Cloud has tried to index their skills and possibly look at career paths or other attributes. Compared to the data resident in Eightfold (over a billion user records), Seekout (nearly a billion), and systems like Retrain.ai, Skyhive, and sourcing systems like Beamery or Phenom, this is a very small amount of data. At some point Workday is going to have to understand that the HCM AI platforms of today are really global workforce data systems, not just customer data systems. So most of the AI well see in Workday will make your version of Workday run a bit better.

Prism: Workdays Strategy To Consolidate Data

Finally let me mention the growth of Prism Analytics (now referred to as just Prism), Workdays open data platform for analytics and third party data. When the company acquired Platfora the original need was to give Workday customers a place to put non-Workday data. Since the Workday data platform is a proprietary, object-based database, there was no way to directly import data into Workday so the company needed a scalable data platform.

Since then Prism has grown exponentially. Initially positioned as an analytics system (you could put financial data into Prism and cross-correlate it with HR data), it is now a big data platform which companies can use for financial applications, HR applications, and just about anything you want. Its not designed to compete with Google Big Query or Red Shift from AWS (at least not at the moment) but for Workday customers who want to leverage their investment in Workday security and existing applications, its pretty powerful.

One of the customers who spoke at the conference was Fannie Mae, who has more than $4 trillion in mortgages and loans in its risk managed portfolio. They are using Prism along with Workday Financials to manage their complex month-end close and other financial analysis. Last year I met a large bank who was using Prism to manage, price, and analyze complex banking securities with enormous amounts of calculations built in. Because Prism is integrated into the Workday platform, any Prism application can leverage any Workday data object, so its really a Big Data Extension to the Workday platform.

And that leads back to AI. If Sayans vision comes true, the Workday platform could become a place where customers take their transactional data, customer data, and other important business data and correlate it with Workday financial and HCM data, using AI to find patterns and opportunities. While AWS, Google Cloud, and Azure will offer these services too, none of these vendors have any business applications to offer. So part of Workdays AI strategy is to enable companies to build their own AI-enabled apps, implemented through Extend and Orchestrate and fueled with data from Prism.

This is going to be a crowded space. Microsofts new Power Platform Copilot and OpenAI Azure Services also give companies a place (and method) to build enterprise AI apps. And Google will soon likely launch many new AI services as well. But for companies that have invested in Workday as their core Financial or HCM platform, there are going to be new AI apps that wind up in the Workday platform and that drives utilization, revenue (through Extend, Prism, and Orchestrate), and even vertical apps for Workday.

Workdays Position For The Future

In summary, Workday is well positioned for this new technology revolution. I did challenge the management team to consider ChatGPT as a a new conversational front end to the whole system and they agreed that it is on their list of things to look at.

(By the way, the creative solutions coming to HR in Generative AI are going to blow your mind. Ill share more soon.)

For enterprise buyers, Workday remains rock solid. With only a few major competitors to think about (Oracle, SAP, UKG, Darwinbox, ADP), the company is likely to continue to grow market share for large companies. There will be pricing pressure because of the economy, but for companies that want a first-class technology platform for core Finance and HR, Workday will continue to be a leader.

Additional Resources

The Role Of Generative AI And Large Language Models in HR

New MIT Research Shows Spectacular Increase In White Collar Productivity From ChatGPT

LinkedIn Announces Generative AI Features For Career, Hiring, and Learning

Microsoft Launches OpenAI CoPilots For Dynamics Apps And The Enterprise.

Understanding Chat-GPT, And Why Its Even Bigger Than You Think (*updated)

Microsofts Massive Upgrade: OpenAI CoPilot For Entire MS 365 Suite.

Originally posted here:
Workday's Response To AI and Machine Learning: Moving Faster Than Ever - Josh Bersin

In 2020, artificial intelligence company OpenAI stunned the tech world with its GPT-3 machine learning algorithm. After ingesting a broad slice of the internet, GPT-3 could generate writing that was hard to distinguish from text authored by a person, do basic math, write code, and even whip up simple web pages.

OpenAI followed up GPT-3 with more specialized algorithms that could seed new products, like an AI called Codex to help developers write code and the wildly popular (and controversial) image-generator DALL-E 2. Then late last year, the company upgraded GPT-3 and dropped a viral chatbot called ChatGPTby far, its biggest hit yet.

Now, a rush of competitors is battling it out in the nascent generative AI space, from new startups flush with cash to venerable tech giants like Google. Billions of dollars are flowing into the industry, including a $10-billion follow-up investment by Microsoft into OpenAI.

This week, after months of rather over-the-top speculation, OpenAIs GPT-3 sequel, GPT-4, officially launched. In a blog post, interviews, and two reports (here and here), OpenAI said GPT-4 is better than GPT-3 in nearly every way.

GPT-4 is multimodal, which is a fancy way of saying it was trained on both images and text and can identify, describe, and riff on whats in an image using natural language. OpenAI said the algorithms output is higher quality, more accurate, and less prone to bizarre or toxic outbursts than prior versions. It also outperformed the upgraded GPT-3 (called GPT 3.5) on a slew of standardized tests, placing among the top 10 percent of human test-takers on the bar licensing exam for lawyers and scoring either a 4 or a 5 on 13 out of 15 college-level advanced placement (AP) exams for high school students.

To show off its multimodal abilitieswhich have yet to be offered more widely as the company evaluates them for misuseOpenAI president Greg Brockman sketched a schematic of a website on a pad of paper during a developer demo. He took a photo and asked GPT-4 to create a webpage from the image. In seconds, the algorithm generated and implemented code for a working website. In another example, described by The New York Times, the algorithm suggested meals based on an image of food in a refrigerator.

The company also outlined its work to reduce risk inherent in models like GPT-4. Notably, the raw algorithm was complete last August. OpenAI spent eight months working to improve the model and rein in its excesses.

Much of this work was accomplished by teams of experts poking and prodding the algorithm and giving feedback, which was then used to refine the model with reinforcement learning. The version launched this week is an improvement on the raw version from last August, but OpenAI admits it still exhibits known weaknesses of large language models, including algorithmic bias and an unreliable grasp of the facts.

By this account, GPT-4 is a big improvement technically and makes progress mitigating, but not solving, familiar risks. In contrast to prior releases, however, well largely have to take OpenAIs word for it. Citing an increasingly competitive landscape and the safety implications of large-scale models like GPT-4, the company opted to withhold specifics about how GPT-4 was made, including model size and architecture, computing resources used in training, what was included in its training dataset, and how it was trained.

Ilya Sutskever, chief technology officer and cofounder at OpenAI, told The Verge it took pretty much all of OpenAI working together for a very long time to produce this thing and lots of other companies would like to do the same thing. He went on to suggest that as the models grow more powerful, the potential for abuse and harm makes open-sourcing them a dangerous proposition. But this is hotly debated among experts in the field, and some pointed out the decision to withhold so much runs counter to OpenAIs stated values when it was founded as a nonprofit. (OpenAI reorganized as a capped-profit company in 2019.)

The algorithms full capabilities and drawbacks may not become apparent until access widens further and more people test (and stress) it out. Before reining it in, Microsofts Bing chatbot caused an uproar as users pushed it into bizarre, unsettling exchanges.

Overall, the technology is quite impressivelike its predecessorsbut also, despite the hype, more iterative than GPT-3. With the exception of its new image-analyzing skills, most abilities highlighted by OpenAI are improvements and refinements of older algorithms. Not even access to GPT-4 is novel. Microsoft revealed this week that it secretly used GPT-4 to power its Bing chatbot, which had recorded some 45 million chats as of March 8.

While GPT-4 may not to be the step change some predicted, the scale of its deployment almost certainly will be.

GPT-3 was a stunning research algorithm that wowed tech geeks and made headlines; GPT-4 is a far more polished algorithm thats about to be rolled out to millions of people in familiar settings like search bars, Word docs, and LinkedIn profiles.

In addition to its Bing chatbot, Microsoft announced plans to offer services powered by GPT-4 in LinkedIn Premium and Office 365. These will be limited rollouts at first, but as each iteration is refined in response to feedback, Microsoft could offer them to the hundreds of millions of people using their products. (Earlier this year, the free version of ChatGPT hit 100 million users faster than any app in history.)

Its not only Microsoft layering generative AI into widely used software.

Google said this week it plans to weave generative algorithms into its own productivity softwarelike Gmail and Google Docs, Slides, and Sheetsand will offer developers API access to PaLM, a GPT-4 competitor, so they can build their own apps on top of it. Other models are coming too. Facebook recently gave researchers access to its open-source LLaMa modelit was later leaked onlinewhile a Google-backed startup, Anthropic, and Chinas tech giant Baidu rolled out their own chatbots, Claude and Ernie, this week.

As models like GPT-4 make their way into products, they can be updated behind the scenes at will. OpenAI and Microsoft continually tweaked ChatGPT and Bing as feedback rolled in. ChatGPT Plus users (a $20/month subscription) were granted access to GPT-4 at launch.

Its easy to imagine GPT-5 and other future models slotting into the ecosystem being built now as simply, and invisibly, as a smartphone operating system that upgrades overnight.

If theres anything weve learned in recent years, its that scale reveals all.

Its hard to predict how new tech will succeed or fail until it makes contact with a broad slice of society. The next months may bring more examples of algorithms revealing new abilities and breaking or being broken, as their makers scramble to keep pace.

Safety is not a binary thing; it is a process, Sutskever told MIT Technology Review. Things get complicated any time you reach a level of new capabilities. A lot of these capabilities are now quite well understood, but Im sure that some will still be surprising.

Longer term, when the novelty wears off, bigger questions may loom.

The industry is throwing spaghetti at the wall to see what sticks. But its not clear generative AI is usefulor appropriatein every instance. Chatbots in search, for example, may not outperform older approaches until theyve proven to be far more reliable than they are today. And the cost of running generative AI, particularly at scale, is daunting. Can companies keep expenses under control, and will users find products compelling enough to vindicate the cost?

Also, the fact that GPT-4 makes progress on but hasnt solved the best-known weaknesses of these models should give us pause. Some prominent AI experts believe these shortcomings are inherent to the current deep learning approach and wont be solved without fundamental breakthroughs.

Factual missteps and biased or toxic responses in a fraction of interactions are less impactful when numbers are small. But on a scale of hundreds of millions or more, even less than a percent equates to a big number.

LLMs are best used when the errors and hallucinations are not high impact, Matthew Lodge, the CEO of Diffblue, recently told IEEE Spectrum. Indeed, companies are appending disclaimers warning users not to rely on them too muchlike keeping your hands on the steering wheel of that Tesla.

Its clear the industry is eager to keep the experiment going though. And so, hands on the wheel (one hopes), millions of people may soon begin churning out presentation slides, emails, and websites in a jiffy, as the new crop of AI sidekicks arrives in force.

Image Credit:Luke Jones /Unsplash

Original post:
OpenAI Says GPT-4 Is Better in Nearly Every Way. What Matters More Is Millions Will Use It - Singularity Hub

Artificial intelligence is one of those technical buzzwords that has captivated the press and is being discussed in practically every facet of life.

A subset of artificial intelligence known as machine learning, in particular, may assist building and construction fleet operators in optimizing fleet performance while preserving safety as a primary concern.

As roads become increasingly dangerous, safety directors face a challenging task: sifting through terabytes of information to uncover weaknesses in fleet safety.

As data analytics technologies develop, safety managers must learn to use them to surf the data tsunami, stay afloat, and protect the safety of their drivers.

Artificial intelligence (AI) is sometimes mistaken for machine learning (ML). People frequently confuse or use the phrases interchangeably, although they are not the same thing.

AI entails having a machine perform something that a person would ordinarily do. Machine learning, on the other hand, refers to technology that learns or works out how to accomplish something on its own.

Read on to learn more about machine learning and how it improves fleet safety.

Giving computers data and letting them learn for themselves is exactly what machine learning entails. This system can take a massive quantity of information and develop models based on behavioral patterns. They can learn and improve themselves autonomously by studying known occurrences without being trained. As the system develops, it may be fed fresh data and used to forecast the outcomes of future occurrences.

Machine learning in fleet management enhances how data analytics systems handle large amounts of data. The system begins to learn which data is most frequently reviewed throughout daily use and adapts itself in real time based on the owners behavior.

Machine learning algorithms enable the building of dashboards with features that make it simple to examine different data points such as vehicle downtime and specific driver behavior that should be rectified. These advanced neural networks can warn drivers when their vehicles need repair or are going to experience mechanical troubles.

Smart fleet management systems are better capable of diagnosing problems than traditional ones. A company can make better overall judgments when it can view the large picture of its fleet in one location. ML and AI technologies can drastically reduce total expenses.

Machine learning is beneficial in all aspects of fleet vehicle management, including efficiency and safety. Manual methods made fleet management laborious and difficult, but machine learning helps simplify operational operations, making them easy and simple.

Moreover, when integrated with machine vision, ML may improve fleet management even further.

Listed below are a few advantages of ML-driven fleet management:

In fleet management, machine learning employs predictive analysis to avoid probable accidents and notify at-risk drivers. A large and full set of historical data may be used to develop a prediction model. It entails examining the actions that led to the accidents.

Using the correct machine learning technology enables risk minimization, accident avoidance, and insurance claim reduction. Understanding, selecting, and then executing the best solution is important to make an accurate and effective forecast.

Integrating centralized data management software with specialized tools allows for rapid information collecting, prediction, and exception handling.

Listed below are a few ways how ML improves fleet safety.

Machine learning has enabled fleet managers to become more proactive in risk management. Early safety measures were quite reactive. They may have detected things like harsh braking circumstances, but they lacked information about what caused them.

Machine learning goes beyond merely logging such occurrences. It may, for example, include a video to demonstrate what caused the brakes. This reduces false alarms and enables fleets to allocate precious resources where they are most needed.

From time to time, police reports are not always reliable after an accident. When cameras were first employed, it became clear that most of the information in police reports was incorrect. When cameras were put at various transportation businesses, it wasnt long before the technology vindicated a driver who had been wrongfully accused of causing a crash by law police.

Fleets that use the most up-to-date safety ML technologies can be more focused in their training initiatives. This has aided driver turnover. Rather than retraining all drivers in the fleet on backing owing to a rise in backing events, the fleet may concentrate on only those drivers whose conduct puts them at a higher risk of backing issues.

Fleet companies that use ML safety solutions must also keep the human factor in mind. This should be considered as another tool in the toolkit for assisting drivers in sharing the road safely.

People do not trust a corporation that goes too far down the ML route and just tells them to absolutely believe this machine and do exactly what it says. It must be included in a program. ML exists to help humans make the best decisions possible.

Data obtained by safety technology also provides a safety manager with confidence when dealing with driving behaviors. They are not making decisions based on assumptions or insufficient information. Driver scorecards gamified performance, creating a competitive environment that drove drivers to perform at their best.

Fleet management is a critical component of running a profitable business. Since crashes are statistically uncommon, and carriers cant look at that fine level of data, one cant make inferences based on what happens in an hour, a day, or even a week. The safety of a fleet may be dramatically enhanced with excellent machine learning in place.

No, machine learning is not the answer to all the problems in fleet management. It can provide various information to the company with which they can solve problems using a variety of methods.

Listed below are some of the factors that can affect machine learning;

Machine learning training necessitates the storage system reading and rereading whole data sets, generally at random. This implies that archive systems that only provide sequential access techniques, such as tape, cannot be used.

You might also like

View original post here:
How Machine Learning Helps Improve Fleet Safety - Robotics and Automation News