AI News This Week: Infrastructure, Innovation, and Education

UWCISA Blog · AI on the Edge: Infrastructure, Innovation, and Education

📰 AI Overbuild: China’s Risky Bet on Data Center Expansion

China has poured billions into building data centers to fuel its AI ambitions, but a new report reveals that up to 80% of this capacity is currently unused. Driven by government initiatives and private investment, more than 500 centers were launched, though many are now idle due to shifting AI demands and poor infrastructure planning. Technologies like DeepSeek have pivoted AI interest from model training to inference, making much of the infrastructure obsolete. Developers are resorting to giveaways and GPU selloffs, raising fears that a wave of excess capacity could disrupt the global data center market.

Source: TechRadar

• Idle Infrastructure: Up to 80% of China’s new data center capacity is currently unused.
• Shift in AI Focus: Tools like DeepSeek’s R1 model have moved the emphasis from training to inference, rendering many training-oriented data centers obsolete.
• Market Impact Risk: If unused capacity hits the general market, it could significantly disrupt prices and investor confidence.

🧠 Agentic vs. Generative AI: What’s the Real Difference?

As artificial intelligence continues to evolve, two distinct forms — Generative AI (Gen AI) and Agentic AI — are shaping the future of automation and decision-making. Gen AI is designed to create original content like text, images, and code based on user input, using deep learning models that emulate human language and logic. In contrast, Agentic AI represents the next leap forward: AI systems capable of acting autonomously, making decisions, and pursuing complex goals with minimal human intervention. These systems blend large language models with machine learning, reinforcement learning, and context awareness to perceive, reason, act, and learn independently.

The article emphasizes that Gen AI is reactive, producing content when prompted, while Agentic AI is proactive, capable of adjusting to its environment and taking initiative. Use cases for Gen AI include SEO content creation, customer support chatbots, and product design, whereas Agentic AI is being tested in sectors like logistics, finance, healthcare, and even city planning. Despite its promise, Agentic AI remains largely experimental, and the distinction between it and traditional AI agents is important: Agentic AI is the overarching framework, while AI agents are the functional components executing tasks within it.

Source: IBM

• Core Difference: Generative AI creates content; Agentic AI autonomously makes decisions and performs tasks with limited oversight.
• Autonomous Capabilities: Agentic AI uses perception, reasoning, and action loops to operate proactively in dynamic environments.
• Emerging Applications: While Gen AI is widely adopted, Agentic AI is being explored in complex domains like urban planning, finance, and robotics.

💡 Microsoft’s Copilot Gets Smarter: Memory, Vision, and a Touch of Clippy

To celebrate its 50th anniversary, Microsoft has rolled out a sweeping update to its Copilot AI assistant, significantly enhancing its capabilities and personalization features. Built on OpenAI models, the updated Copilot now includes Memory, allowing it to retain user preferences, interests, and even personal details like birthdays to offer more tailored suggestions. Users can control what Copilot remembers or choose to opt out.

New additions like Actions enable the AI to interact with websites directly—booking tickets, making purchases, or navigating the web—similar to tools like OpenAI's Operator. Copilot Vision, previously limited to web tools, now expands to Windows and mobile platforms, offering real-time screen and camera analysis to answer questions or help manage documents. Another standout feature is Deep Research, which lets Copilot analyze massive datasets or documents to support complex projects and integrates tightly with Bing search.

Microsoft also introduced podcast generation capabilities and a new Pages feature to help organize information across files. While many of these tools mirror capabilities already available in competing AI platforms, Microsoft’s decision to launch them all at once demonstrates a robust effort to stay competitive and to deliver a deeply personalized AI experience. Oh, and yes—Clippy might be making a comeback.

Source: The Verge

• Major Feature Expansion: Microsoft Copilot now supports memory, personalization, web actions, podcast creation, and document analysis.
• User-Centric Customization: Copilot can remember user preferences and soon offer a personalized visual interface—possibly including Clippy.
• Competitive Push: With many features mirroring rivals like ChatGPT and Google Gemini, Microsoft is aggressively enhancing Copilot to maintain AI leadership.

🦙 Meta Unleashes LLaMA 4—But the Real AI Powerhouse Is Yet to Come

Meta has officially unveiled its latest suite of large language models, LLaMA 4, as part of its ongoing effort to stay competitive in the generative AI space. While the new models are available now to developers and researchers, CEO Mark Zuckerberg emphasized that Meta’s most powerful AI model is still in development, hinting at an even more advanced system expected to launch later in 2025. LLaMA 4 builds on the success of its predecessor by enhancing performance, safety, and scalability, with multiple versions designed to meet varied application needs.

A key distinction in Meta’s approach is its open-source strategy, which aims to foster community collaboration and innovation—contrasting sharply with competitors like OpenAI and Anthropic, who have kept their most powerful models proprietary. Meta plans to integrate LLaMA 4 into its suite of products, including WhatsApp, Instagram, and Facebook, and even roll out AI agents that users can interact with directly in-app. Additionally, Meta is laying the groundwork for a public AI infrastructure, allowing businesses to build their own applications atop Meta’s models through APIs and developer tools.

Zuckerberg positioned this latest release not as an endpoint, but as a stepping stone toward a future where AI is deeply embedded across consumer and enterprise tools. With one eye on rival advancements and the other on a larger vision for democratized AI access, Meta is staking a major claim in the AI landscape.

Source: CNBC

• LLaMA 4 Released: Meta launched new versions of its open-source AI models, with broader integration planned across its apps.
• More to Come: CEO Mark Zuckerberg revealed Meta’s most powerful model is still in development and slated for a 2025 release.
• Open Source Focus: Meta continues to differentiate itself with a transparent, community-first approach to AI development.

🏆 Microsoft Launches Global AI Skills Fest with Record-Breaking Ambitions

In a creative and ambitious push to both democratize AI education and set a Guinness World Record, Microsoft has launched its AI Skills Fest—a 50-day global training initiative offering free AI education to learners of all levels. The event aims to have over 46,046 participants complete a multi-level AI lesson within 24 hours between April 7 and 8, 2025, breaking the current record for most users completing an online AI course in a day. The fest includes a mix of beginner-friendly tutorials and more advanced modules on topics like machine learning, natural language processing, and Microsoft’s Azure and Copilot tools.

The training is available in over 30 languages and includes everything from self-paced lessons and hackathons to sweepstakes and discount codes. Microsoft has gamified the experience with prizes like certification vouchers and even a LinkedIn-eligible participation badge. While it may have a playful marketing tone, the initiative also serves a broader mission: to equip more people with AI literacy at a time when demand for these skills is soaring. GitHub, a Microsoft subsidiary, is also participating by offering discounts on its Copilot certification for developers.

With both accessibility and fun at the forefront, AI Skills Fest not only promotes education but turns learning into a global event—blending practical upskilling with a touch of spectacle.

Source: ZDNet

• AI for All: Microsoft’s AI Skills Fest offers free training for all experience levels in a wide variety of AI topics and tools.
• Guinness World Record Attempt: The company hopes to break the record for the most users completing an AI course in 24 hours.
• Gamified Learning: Participants can earn badges, win certification vouchers, and join community events while gaining practical AI skills.

Five Top Tech Takeaways: Nvidia's Plunge, Claude 3.5, Napster-esque Legal Battles and McDonald's Ends AI Experiment

AI Sell-off?

Nvidia's AI-Fueled Surge Halts with 13% Stock Plunge

Source: CNN

Nvidia has lost its position as the world's most valuable listed company after a significant stock decline. The company's market capitalization fell from $3.34 trillion to $2.91 trillion in the past week, dropping Nvidia to third place behind Microsoft and Apple. The stock tumble is attributed to investor profit-taking and a cooling excitement over Nvidia's role in the AI revolution. Despite this setback, Nvidia's stock had surged 139% over the past year, fueled by its AI-powered chips. The broader market remained relatively stable, with gains in sectors like energy and financials balancing out Nvidia's decline.

• Stock Decline: Nvidia's market cap dropped by $430 billion, leading to its fall to third place among the world's most valuable companies.
• Investor Sentiment: The decline reflects typical volatility and profit-taking after significant gains, with a broader market showing stability.
• AI Hype: Nvidia's role in AI has driven substantial stock gains, but recent drops indicate a possible cooling of investor enthusiasm.

Claude 3.5 Sonnet: Anthropic’s Latest AI Model Outshines Competitors

Source: The Verge

Anthropic has introduced its latest AI model, Claude 3.5 Sonnet, positioning it as a strong competitor to OpenAI's GPT-4o and Google's Gemini. The new model is already available to users and developers, outperforming its predecessor, Claude 3 Opus, in various benchmarks. Claude 3.5 Sonnet excels in writing and translating code, handling complex workflows, interpreting charts, and transcribing text from images. Additionally, the model introduces a new feature called Artifacts, allowing users to interact and edit AI-generated content within the app. Anthropic aims to transform Claude into a comprehensive tool for businesses to manage knowledge and documents securely.

• New AI Model: Claude 3.5 Sonnet is launched, outperforming its predecessor and major competitors in several benchmarks.
• Enhanced Features: The model excels in multiple tasks, including coding, workflow management, and text transcription, with better understanding of humor and human-like writing.
• Artifacts Feature: Introduces interactive editing of AI-generated content within the app, signaling a move towards comprehensive business solutions.

Apple and Meta: Rivals to Collaborators in AI Push

Source: The Wall Street Journal

Apple is in talks with Meta to integrate Meta’s generative AI model into Apple Intelligence, the new AI system for iPhones and other devices. This collaboration, if finalized, would allow Meta to leverage Apple’s vast distribution network. Apple has already partnered with OpenAI and is exploring similar deals with Google, Anthropic, and Perplexity. The potential partnership between Apple and Meta marks a significant shift given their historical rivalry. Apple aims to offer multiple AI options to users, enabling personalized choices for various tasks. This strategic move helps Apple diversify its AI offerings and reduces reliance on a single partner.

• Potential Partnership: Apple and Meta are discussing integrating Meta’s AI model into Apple Intelligence, leveraging Apple's distribution.
• Diversification Strategy: Apple seeks multiple AI partnerships, including with OpenAI, Google, Anthropic, and Perplexity, to offer diverse AI capabilities.
• Historic Rivalry: The collaboration represents a notable shift in the relationship between Apple and Meta, historically competitors in the tech space.

Legal Battle Erupts Between Music Labels and AI Startups

Source: Yahoo Finance

Major record companies, including Sony Music Entertainment, Universal Music Group, and Warner Records, have filed lawsuits against AI music startups Suno and Udio for copyright infringement. The lawsuits, filed in federal courts in Boston and New York, allege that these AI song-generators are exploiting the works of artists without permission. Suno AI's CEO, Mikey Shulman, denies the claims, stating their technology generates new outputs without memorizing existing content. This legal action highlights ongoing tensions in the music industry regarding the use of AI, with concerns about legality and protecting artists' rights.

• Lawsuits Filed: Major record labels are suing AI startups Suno and Udio for alleged copyright infringement.
• Technology Defense: Suno AI claims its technology generates new outputs and does not replicate existing content.
• Industry Tensions: The lawsuits reflect broader concerns in the music industry about AI's impact on artists' rights and the legality of AI-generated content.

Fast Food Giant McDonald's Stops AI Drive-Thru Experiment

Source: The Guardian

McDonald's is discontinuing its test of AI chatbots in drive-thrus across over 100 US locations, ending a trial that began in 2021 with IBM. Despite no public reason given for the shutdown, McDonald's confirmed that a voice-ordering solution remains a future goal. The fast food industry has shown a strong interest in AI, with chains like Wendy's and Taco Bell also implementing AI technologies to reduce labor costs. However, AI's reliability has been questioned, with viral videos showcasing order inaccuracies. Additionally, the industry faces criticism for relying on outsourced labor to support AI systems.

• AI Test Ended: McDonald's ends its AI chatbot drive-thru test, impacting over 100 locations in the US.
• Future AI Plans: McDonald's and other fast food chains continue to explore AI solutions despite mixed results and reliability issues.
• Industry Trend: Fast food chains are increasingly adopting AI to cut labor costs, despite challenges with accuracy and reliance on outsourced labor.

Author: Malik Datardina, CPA, CA, CISA. Malik works at Auvenir as a GRC Strategist who is working to transform the engagement experience for accounting firms and their clients. The opinions expressed here do not necessarily represent UWCISA, UW, Auvenir (or its affiliates), CPA Canada or anyone else. This post was written with the assistance of an AI language model. The model provided suggestions and completions to help me write, but the final content and opinions are my own.

Five Top Tech Takeaways: Lawyer/GenAI Trouble (Again), FTX Insider on the Stand, Coffee Badging Explored, Adobe Strikes Back, and Quantum Computing in Canada

Robot Lawyer Getting Grilled

Generative AI on the Stand (Again): Legal Outcomes Questioned in High-Profile Case

Prakazrel “Pras” Michel, a member of the Fugees, has challenged his conviction, citing his defense's reliance on generative AI for closing arguments as a critical failure. This unprecedented use of AI, touted as a game-changer for complex litigation, backfired, with attorney David Kenner allegedly confusing case details and misquoting lyrics. While the tool's developer, EyeLevel.AI, defends its application, emphasizing its role in enhancing legal procedures rather than replacing human input, critics argue the technology's infancy and propensity for generating “hallucinations” or misleading information. The situation underscores the legal industry's rapid and potentially precarious engagement with AI, as experts warn of the need for meticulous review standards. As law firms increasingly adopt AI, this case serves as a cautionary tale, highlighting the technology's potential and pitfalls in critical legal scenarios. (Source: Yahoo Finance)

The incident serves as a good example of "abdication" when using generative AI. Professionals should instead focus on use cases that look at amplifying one's effort instead. For more, check out my post on Medium, which also includes links to previous mishaps involving lawyers, fake cases, and ChatGPT. 

From AI to Quantum: Canadian Tech Takes a Giant Leap

Bromont, Quebec, hosts one of the world's few quantum computers, specifically the IBM Quantum System One, marking a monumental stride in both scientific and commercial sectors. This exclusive installation, the only one in Canada, underscores the location's rich talent pool and expertise, particularly in artificial intelligence and quantum computing convergence. The groundbreaking technology promises to revolutionize computations, tackling more complex, multi-variable calculations unreachable by current standards, primarily benefiting logistics, sustainable development, and potentially the financial sector. While still in the exploratory phase with certain operational challenges, quantum computing is poised for substantial evolution, anticipating growth from its current 127 qubits to over a thousand, significantly minimizing error rates and enhancing computational longevity and complexity. This advancement not only cements Canada's position on the global tech stage but also beckons foreign enterprises, asserting its status as an emerging, influential hub in technology and innovation. (Source: CTV News)

FTX Unraveled: Singh’s Testimony Highlights Billion-Dollar Discrepancies

In a recent development that has rocked the financial and tech sectors, Nishad Singh, a former executive and close associate of Sam Bankman-Fried (SBF), founder of FTX, has made serious allegations in court, suggesting high-level financial misconduct within the company. Detailing his interactions with Bankman-Fried, Singh's testimony provides an insider's view of a series of alleged fraudulent activities and fiscal mismanagement that led to significant financial discrepancies within the firm.

Specific allegations by Singh include:

• Misappropriation of FTX customer funds by Alameda, another company under Bankman-Fried.
• A staggering $13 billion debt from Alameda to FTX, with an inability to repay $8 billion of this amount.
• Alameda was operating with insufficient collateral, often falling short by as much as $10 billion.
• Unchecked excessive spending by Bankman-Fried, including lavish investments and endorsement deals totaling over $1 billion.
• Investments into ventures perceived as non-beneficial and toxic to FTX and Alameda's culture.
• Manipulation of financial data, such as creating back-dated transactions to falsify FTX's revenue figures.
• Use of a code feature, “allow_negative,” for inappropriate internal transactions.
• Singh's involvement in shady transactions at Bankman-Fried's behest, including transfers to obscure accounts and back-dating token transfers to deceive regulatory bodies.
• Political donations made from Singh's accounts without his direct consent, highlighting a lack of financial control among company executives.
• Emotional and mental duress leading to suicidal ideation for Singh during the company’s collapse.
• Singh’s candid courtroom revelations depict a scenario of betrayal, intimidation, and a flagrant abuse of power and trust within the upper echelons of FTX, raising grave concerns about corporate governance in rapidly growing tech companies.

 As noted in The Verge's original piece, Singh was one of many to turn on SBF: "On November 20th, 2022, Singh had his first meeting with prosecutors. Of the inner circle, Wang cooperated first. Singh was second. Ellison was third." Source: The Verge)

"Coffee Badging": A Clever Hack or a Career Hazard?

"Coffee badging" is a term that has gained traction in the corporate world, referring to the practice of employees briefly appearing at the office, only to leave shortly after. As companies increasingly require employees to return to the office, some workers, rather than defying these mandates, have chosen to abide by them in the most minimal way possible. They may appear on days when few colleagues are around, leave early, or barely show up at all. This is akin to students signing attendance sheets and then skipping lectures. While it may seem like a clever hack to the return-to-office system, there are potential pitfalls. Keith Styrcula, CEO of Glasstower Digital, notes that face-to-face time and office politics can be as crucial as merit in the corporate world. A recent survey by KPMG revealed that nearly two-thirds of U.S. CEOs expect employees to work full-time in offices within three years, and 90% would reward those who frequent the office with favorable assignments and promotions.

Key Takeaways:

• "Coffee badging" refers to the act of briefly showing up at the office and leaving soon after.
• While it may seem advantageous in the short term, long-term consequences could include missed opportunities and promotions.
• A KPMG survey found "almost two-thirds said they envision employees working in offices full-time three years from now, and 90% said they’ll reward office goers with favorable assignments, raises and promotions. ".

Source: The Wall Street Journal

AI-Powered Video Editing: Adobe Strikes Back

Adobe recently showcased its innovative generative fill feature named "Project Fast Fill" at the MAX conference. This feature, powered by artificial intelligence, can effortlessly add or remove objects in videos. It offers capabilities like changing clothing accessories on moving people or erasing tourists from video landscapes. This new feature operates similarly to Google’s Magic Editor but is designed for videos instead of still images. With the advancement of Adobe’s updated Firefly AI models, Project Fast Fill can also modify colors in still photos by merely inputting a text prompt. Although currently experimental, past trends indicate that such features eventually become available to Creative Cloud users.

Key Takeaways:

• Adobe's "Project Fast Fill" allows users to add or remove objects in videos using AI.
• Adobe is striking back at its competition: the feature works similarly to Google's Magic Editor but is designed for video content.
• Adobe is expanding its AI editing technology to other areas like video, audio, and 3D design, introducing features like voice translation and 3D object posing

(Source: The Verge)

Author: Malik Datardina, CPA, CA, CISA. Malik works at Auvenir as a GRC Strategist who is working to transform the engagement experience for accounting firms and their clients. The opinions expressed here do not necessarily represent UWCISA, UW, Auvenir (or its affiliates), CPA Canada or anyone else. This post was written with the assistance of an AI language model. The model provided suggestions and completions to help me write, but the final content and opinions are my own.

AI and the CPA: OpenAI's ChatGPT will Blow Your Mind! Try it Now!

We previously discussed OpenAI in this post. In that post, I closed with this comment:

"We should be aware of the concept of exponential change. Technology, like AI, improves at an exponential rate and not a linear rate. Consequently, monitoring the space is key for CPAs and other to ensure that they see change coming and adapt accordingly."

And so, my fellow CPAs, you need to check out how much AI has advance in the two years since that post. It's absolutely amazing to see how advanced the technology is and how quickly it has come:

YouTube is one thing. But testing it out is another. To be honest, I was about to review the video and then summarize it. Then I thought, why not ask ChatGPT? 

Question to ChatGPT: "What are 5 key takeaways for CPAs with respect to ChatGPT?"

Answer: 

1. ChatGPT is a large language model trained by OpenAI.
2. ChatGPT has a knowledge cutoff of 2021, and it does not have access to the internet or current events.
3. ChatGPT is able to generate human-like responses to a wide range of questions, but it is important to keep in mind that it is a machine and may not always provide accurate or complete information.
4. ChatGPT is not a substitute for professional advice or guidance, and it should not be used as the sole basis for making important decisions.
5. ChatGPT is intended for general informational purposes only, and it is not a substitute for professional judgment or expertise. It is important for CPAs to consult with appropriate sources and to exercise their professional judgment when making decisions.

Reading through the takeaways, it seems like it added an extra dash of risk management. Could it have accounted for our risk-aversion?

Then I thought, why not ask it directly about the video itself? 

Question to ChatGPT: "What are 5 key takeaways from It’s Time to Pay Attention to A.I. (ChatGPT and Beyond)"

Answer:

1. Artificial intelligence (AI) is a rapidly developing field with the potential to significantly impact many aspects of society, including the accounting profession.
2. AI technologies, such as chatbots and language models like ChatGPT, are becoming increasingly sophisticated and can be used for a wide range of applications.
3. It is important for CPAs to stay informed about the latest developments in AI and to consider how these technologies may impact their profession and their clients.
4. While AI has the potential to improve efficiency and accuracy, it is important to carefully consider the risks and limitations of these technologies, including the potential for errors or biases.
5. CPAs should consider ways to incorporate AI into their practices, such as by using AI tools to automate certain tasks or by providing guidance to clients on the use of AI. However, they should also ensure that they have the necessary skills and knowledge to effectively use and understand these technologies.

Notice how I did not have to tailor this questions for CPAs? ChatGPT automatically brought the context over from the previous question.  (As for how good this summary is, you can judge for yourself.)

Then, I decided to ask it about how exponential technologies will impact the CPA profession. 

First question, "Will blockchain replace the need for financial auditors?"

Answer:

"Blockchain technology has the potential to revolutionize many aspects of the financial industry, including auditing. However, it is unlikely that blockchain will completely replace the need for financial auditors.

One of the main features of blockchain is its ability to provide a tamper-evident record of transactions. This can be beneficial for auditing purposes, as it allows auditors to more easily verify the accuracy and completeness of financial records. However, auditors still need to be able to interpret and analyze the data contained in these records, and to ensure that the data is consistent with other information and documentation.

Additionally, while blockchain can provide a secure and reliable record of transactions, it is not a substitute for the expertise and judgment of trained financial professionals. Auditors bring a deep understanding of financial principles and practices, as well as the ability to identify and assess risks and potential areas of concern.

Overall, it is likely that blockchain will augment and enhance the role of financial auditors, rather than replacing them entirely."

Notice the nuanced response? It should have spoke to validity, instead of completeness when it came to how the blockchain will assist the auditor. However, it didn't just list all the audit objectives just "to be safe". 

Then I asked about artificial intelligence:

Question: "How can AI be used to automate financial auditing?"

Answer:

"Artificial intelligence (AI) can be used to automate certain aspects of financial auditing, such as the analysis and interpretation of financial data. This can potentially improve the efficiency and accuracy of the audit process.

For example, AI algorithms can be trained to analyze large amounts of financial data and identify patterns and trends that may not be immediately apparent to human auditors. This can help auditors to more quickly and effectively identify areas of potential risk or concern.

AI can also be used to automate the testing of transactions and the evaluation of internal controls. For example, an AI system could be trained to perform automated tests on transactions to ensure that they are recorded correctly in the financial records. This can help to reduce the time and resources needed to perform these tasks manually.

Overall, the use of AI in financial auditing can help to improve the efficiency and accuracy of the audit process, while also freeing up auditors to focus on more complex and higher-level tasks that require human judgment and expertise."

Recognize the theme of "race with the machine" in both this response and the previous blockchain response? This is especially integrated into the last paragraph, which speaks to what AI can do versus what "human judgment and expertise" can do. Also, notice how it explains AI role in both risk assessment and testing of internal controls. That is, it had enough where "understanding" to breakdown the response into different aspects of the audit. 

When looking at this, we need to ask ourselves: is this inflection point for exponential rise of AI? 

Do review the last part of the video, where Samuel H. Altman, CEO of OpenAI, explains how he anticipates the impact of such technology on the legal profession. (The video earlier takes about how Josh Browder's Do Not Pay, will leverage the tech. I had previously seen Browder on a panel in 2016; see this post for the video). 

It's not quite a stretch to apply what he says to the CPA profession. There are key differences; in that we opine on financial statements, provide tax advice based on financial data, and the like. However, audit data analytics tech has been around for decades. It's just a matter of getting the different parts to talk to each other. 

Clearly, it's early days for ChatGPT and many issues need to be sorted out. For example, it has already earned the moniker "CheatGPT" for how it can be potentially used as a short-cut by students. That being said, it's clearly the biggest watershed moment for AI and the white-collar workforce, since IBM's Watson defeated Ken Jennings and Brad Rutter. 

Author: Malik Datardina, CPA, CA, CISA. Malik works at Auvenir as a GRC Strategist that is working to transform the engagement experience for accounting firms and their clients. The opinions expressed here do not necessarily represent UWCISA, UW, Auvenir (or its affiliates), CPA Canada or anyone else

AI and the CPA: What should CPAs know about GPT3?

In the epic battle between man versus machine, the chess champion Gary Kasparov threw the match because it is alleged that he thought that the AI-powered system did a well-calculated move. In reality, it was just a random move that the system threw out because it had experienced a technical glitch.

One up-and-coming AI-enabled service to watch is GPT-3 from an organization called OpenAI. OpenAI was started by Elon Musk. It’s in beta right now and in a really raw state but the capabilities that have surfaced are pretty amazing. 

One capability (as noted in the video) it has is summarizing long reads. However, there are serious flaws that need to be worked out. For example, it advised a fake patient suffering from depression to kill themselves. So it's not going to be rolled out at a hospital any time soon, but it is something definitely watch.

AI and CPAs: Competitors or Collaborators?

AI could make the profession more sustainable, as these mundane tasks could be handed to a system. MIT’s Eric Brynjolfsson describes this concept as "race with the machine". The idea is that doctors, accountants, lawyers, can work better together with technology. It’s almost like a second set of eyes or someone that can help you assess whether the professional judgement on an issue is correct.

However, this is not something that is currently on the horizon. 

What’s more realistic is understanding where the economics of automation will apply for more basic things like have a more timely close. McKinsey put out a study in August 2020 that found automating and increasing the accuracy of forecasts helped management make better decisions. One case study they highlighted was a manufacturer that was able to reduce inventories and product obsolescence by 20 to 40 percent. 

Change is coming faster than we expect

We should be aware of the concept of exponential change.  Technology, like AI, improves at an exponential rate and not a linear rate. Consequently, monitoring the space is key for CPAs and other to ensure that they see change coming and adapt accordingly. 

Author: Malik Datardina, CPA, CA, CISA. Malik works at Auvenir as a GRC Strategist that is working to transform the engagement experience for accounting firms and their clients. The opinions expressed here do not necessarily represent UWCISA, UW, Auvenir (or its affiliates), CPA Canada or anyone else

Serendipity: Beyond the reach of Robot Professionals?

Came across a story about how Dr. Behfar Ehdaie at Memorial Sloan Kettering Cancer Center was figuring out how to deal with the emotions that come with a discovery of prostate cancer. His novel solution gives may give us some insights into the limits of robots in the professional world.

What he found was that his patients opted for radical treatments, such as surgery or chemotherapy, that resulted in side-effects that actually ended up being more harmful. To use a cliche, the cure was worse than the cancer.

For such patients, "the medical consensus is that active surveillance often is the appropriate treatment for small early tumors". Of course, such an approach is not risk-free, but the problem is that "despite the data showing that this approach is safe, about 50% of eligible men don’t get it either because they turn it down or their physicians don’t embrace it. Medical experts say many men have been overtreated, as their cancers probably posed little immediate danger."

What was his solution?

Negotiate with patients.

As noted in the WSJ article referenced above, he contacted Harvard professor Deepak Malhotra who had authored an article on the topic to develop strategies on negotiating with the patient. Leveraging lessons from behavioral economics was to make monitoring the anchor instead of surgery or chemo. Dr. Ehdaie and professor Malhotra devised a lecture that was delivered to doctors to help them learn from Dr. Ehdaie's successes with this approach.

But what does this have to do with limits of robot-professionals or robopros?

When it comes to cancer treatment and robots, one can't complete the conversation without mentioning IBM's Watson "Oncology Edition". In fact, IBM has a partnership with the same Memorial Sloan Kettering Cancer Center that Dr. Ehdaie works at. Here is a promo-video that speaks to the promise of Watson:

The key to understanding the limits to robot-professionals is the backstory on how Dr. Ehdaie first decided to explore negotiations a way to deal with the issue.

This is where serendipity comes to play.

He was exposed to such concepts with discussions with his wife who is an MBA. Meaning that he went beyond the cancer treatment journals and then discovered a non-standard approach to dealing with a problem. Robots are not good at this. Machine-learning and AI is only good as what you teach it. Even "simple" tasks require thousands of man-hours to train such algos. Perhaps this can be overcome, but currently, it is a real limitation of AI.

Does this make humans indispensable?

Really depends on the objectives that govern the profession and the organizations that hire them.

If it's about cost-cutting and making the process efficient and streamlined, robots are perfect creating a fossilized bureaucracy that is resistant to change. Think about how financial institutions have yet to overhaul their ancient banking systems coded in COBOL:

"In the United States, the financial sector, major corporations, and parts of the federal government still largely rely on it because it underpins powerful systems that were built in the 70s or 80s and never fully replaced."

Similarly, if Dr. Watson replaces a large component of the diagnostic process it would become hard to dislodge it from the cancer treatment process.

On the other hand, if organizations recognize the value of human beings in being important to overall objectives of the profession - patient care, audit quality, etc. - then human judgment must be hardwired into the organization's DNA to avoid the development of such an inflexible system.

Author: Malik Datardina, CPA, CA, CISA. Malik works at Auvenir as a GRC Strategist that is working to transform the engagement experience for accounting firms and their clients. The opinions expressed here do not necessarily represent UWCISA, UW, Auvenir (or its affiliates), CPA Canada or anyone else

RPA and the Accountant: A path out of the mundane?

One of the latest hype technologies is Robot Process Automation (RPA).

My first question when coming across this, is what is the difference between this and cognitive computing? 

As can be seen by these videos, it's more about "dumb" automation instead of "smart" innovation: where routine tasks are handled by the system instead of a person. This is in contrast to something like IBM's Watson, which attempts to understand language and offer probabilistic judgments as to what is the right answer to a question like it did on Jeopardy!


The first video (produced by Deloitte UK) does a great job of actually showing us how RPA can automate the process of extracting information/documents from email and the generating invoices through the company's ERP:

The strength of this video (produced by EY) is showing us the business case for RPA:

The idea is that RPA can automate routine tasks, instead of offshoring. In other words, it brings the world of automation onced reserved for the assembly line to the back office.

As described in this Deloitte publication, it puts RPA as the first step towards a cognitive enterprise - automate the task and then bring cognitive, AI, machine learning, etc., into the process to make it smarter.

To use a maturity model approach, RPA is the first level in bringing together the necessary data and processes to actually train the algorithm to make it smarter.

What does this mean for auditors and accountants?

For accountants, the back office is going to require less people in terms of executing these mundane tasks.

However, this doesn't necessarily mean that jobs will be lost.

As with the advent of cloud computing, the enterprises will have to determine whether such talent can be used more effectively to improve the quality of financial reporting and work on the back log of finance projects that haven't been attended due to staff working on these low-value tasks. That being said, the problem of meeting quarterly targets to feed investors insatiable desire for profits is something that can't be ignored when discussing whether management will choose profits over better processes.

For auditors the story is a little different.

The reality of the profession is that it can't retain talent because people find the work unsustainable: it's hard to shutdown your personal life for a third of the year or more to meet the needs of clients during busy season.

RPA and automation could make the profession more sustainable, as these mundane tasks could be handed to a system instead of a junior. This is similar to the "race with the machine" concept I mentioned in this post, when referring how Watson is helping doctors treat cancer.  Auditor could then focus on more value added tasks, such as assessing aggregate risks, industry trends, etc. Such insights will improve audit quality and give clients better understanding of business and audit risks, making the work more interesting for both auditors and auditees alike.

Author: Malik Datardina, CPA, CA, CISA. Malik works at Auvenir as a GRC Strategist that is working to transform the engagement experience for accounting firms and their clients. The opinions expressed here do not necessarily represent UWCISA, UW, Auvenir (or its affiliates), CPA Canada or anyone else.

CATS2016: Exponential Tech & the CPA

Today, I presented at the Canadian Accounting Technology Show and discussed how exponential technologies and their potential impact on the profession.

During the presentation, I promised a blogpost for the attendees who wanted to dig deeper in the presentation. So here it is!

IBM Watson's Victory over Ken Jennings
During the talk I refer to Ken Jennings and Brad Rutter's defeat at the hands of IBM's Watson. (See Engadet's video for more on this "exponential event".)
This post gives some background on the new "space race" between the tech-giants for the killer AI app and also gives a link to Ken's talk.

For additional information on Watson and the medical profession check out this video.

Exponential versus Linear Technological Change
Kodak - who invented the digital camera in 1975 - was ultimately disrupted by that very same technology. In fact, one of their employees applied Moore's Law to pixel's per dollar in digital cameras.

Why?

The problem illustrate that Kodak (as well as Polaroid) had linear thinking and didn't realize how quick digital technology would become the norm and preferred way of consuming photography. In this post, Peter Diamandis talks about how 30 exponential steps contrasts to 30 exponential steps (and talks more broadly about linear vs exponential thinking) and Ray Kurzweil talks about the infamous story of how the inventor of chess requested an exponential amount of rice (and is rumoured to have lost his head).

Predictions on the Automation of White Collar Work:
These stats are what actually prompted me to propose to CPA Canada that we should have a talk that would discuss this phenomenon. The variety of sources that have chimed in on the topic - combined with the understanding of exponential change - highlights the importance of looking deeper into the trend instead of dismissing it as just fear, uncertainty and doubt (FUD). This of course is not just limited to the accounting profession, but impacts all white collar worker (check out IBM's Watson latest application to automate aspects of the legal profession

• "Job destruction will happen at a faster pace, with machine-driven job elimination overwhelming the market's ability to create valuable new ones.” (Gartner)
• “…knowledge work automaton tools and systems could take on tasks that would be equal to the output of 110 million to 140 million full-time equivalents (FTEs).”’ (McKinsey)
• ‘94% probability accounting/auditing will be automated’ (Oxford Study)
• Finance Department has seen a decrease from an average of 119 people (2004) to 71 people (2014); a reduction by 40% (Hackett Group; as taken from this WSJ article "The New Bookkeeper Is a Robot")

Exponential Technologies
As noted during the presentation, the key exponential technologies that are likely to enable the automation.

Artificial Intelligence: "Science of making computers do things that require intelligence when done by humans." During the presentation, I mention this pharmacist robot being able to dramatically reduce medications errors, which according to the FDA is responsible for 1.3 million injuries.

For other information check out this Deloitte publication on AI and Cognitive.

Internet of Things: "Billions of interconnected sensors and devices will soon exchange data; effectively the physical flow of goods, people, and things will now leave a “digital trail”." RFID inventory does provide some insights in how this digital exhaust left by physical goods can improve inventory management and responsiveness to customers (see this RFID Journal article for more details).

For more on IoT, check out the Deloitte TMT Prediction regarding the technology.

Blockchain: "The blockchain dis-intermediates the need for a centralized trusted authority to administer an exchange of value between parties." As I note in the presentation, I feel the blockchain needs a lot of nuance when discussing how the technology has the potential to disrupt the profession. The technology (as implemented in the exchange of the cryptocurrency Bitcoin) itself won't replace the audit because its controls are designed for the purposes of giving comfort to a retailer, such as Overstock.com, that the buyer has not spent the currency somewhere else. However, if a retailer was then to tell an auditor that they sold goods to these public addresses, the auditor would need to verify that the retailer was not selling the goods to itself (i.e. they would need to verify that the addresses that the retailer sold to are not controlled by the retailer). In other words a sale for the purposes of Bitcoin is not a sale for accounting purposes.

That being said, auditors can’t ignore blockchain as it is the first decentralized approach to exchange value that eliminates the need for a trusted intermediary.

To understand the blockchain better, check out the following videos:

• Blockchain technology will drastically change our lives: This video gives a good overview of the implications of bitcoin and illustrates the role of the network in maintaining the ledger.
• How Bitcoin works under the hood: There is a 5 minute non-tech video, 5 minute tech video and a 22 minute video, which all do a good job of using animation to explain how bitcoin is tamper-proof.
• Khan Academy: The videos are about 90 minutes in total, but it is comprehensive. 

Crowdsourcing: "Process of obtaining needed services, ideas, or content by soliciting contributions from a large group of people, and especially from an online community, rather than from traditional employees or suppliers."

For more on crowdsourcing, I wrote a post on the potential impact on crowdsourcing. The post gives a good background exploring the use-cases brought up by Jeff Howe (who coined the term crowdsourcing).

Near the end of the post, I noted that:

"Can accountants/auditors be crowdsourced like the way professional photographers were? It seems were crowdsourcing works best is an arena where you find hobbyists who do such things out of passion instead of obligation."

Since writing that post I found Gigwalk which illustrates how non-expert tasks within accounting or auditing can be done by the crowd (see this post near the bottom). Also, during the CATS conference it was noted that 50% of practitioners will be retiring over the next 5 to 10 years. Such retirees could form a huge pool of people who want to work casually in their retirement thereby enabling the audit to be crowdsourced.

Concluding thoughts
To meet the challenge of the exponential change, I feel that we need to do the following:

• Hands-on Approach to Technology: University courses on programming, data analytics and data sciences should become a standard part of the accounting student's education. Although tools change over time, I think accounting students who have an open-source statistical package like R would have more options in terms of employment. With respect to data science, (audit) sampling belongs to an era of small data. Consequently, for auditing theory to be keeping pace with the way big data is transforming the way organizations are dealing with their data auditors need to be able traverse data science and auditing theory. 
• Bring in the "hackers": An extension of the above recommendation, is to get the people who think outside the box and disrupt the way we do things.
• Greater focus on cyber security: According to Alec Ross, cyber security is currently a 400 billion dollar problem and is expected to be a $175 billion industry by 2020. Security is a natural extension for CPAs who already need to understand internal controls, governance and concepts of risk (impact, likelihood, threats, etc.). With IoT, the security risks can only be expected to grow exponentially as now even the IoT-enabled fridge can be hacked (and the FTC thinks so as well).
• Smart Contracts+AutoRepos of Smart Cars = Flash Crash¹⁰: As I have written previously about AlgoTrust (second post and first post), I noted that this was another area that CPAs can focus on - auditing algorithms. Just imagine how, these algorithms can feed into blockchain enabled smart contracts that could trigger a massive repossession of smart cars - leaving a city in chaos as people try to figure out how to get home. In other words, CPAs can act as independent monitors of algorithms to ensure such risks are safeguarded against. 
• CPAs-as-a-Crowd: CPAs should leverage the combined power of social and cognitive to get smarter by sharing knowledge and using "smart rooms" that use machine learning and other AI technologies. 

To brings such change the profession, will not the work of one entity alone. Firms, educators, professional bodies and companies need to work together to ensure that the CPA profession will thrive in the world of exponential change that is just around the corner. 

Author: Malik Datardina, CPA, CA, CISA. Malik works at Auvenir as a GRC Strategist that is working to transform the engagement experience for accounting firms and their clients. The opinions expressed here do not necessarily represent UWCISA, UW, Auvenir (or its affiliates), CPA Canada or anyone else.

Cloud vs corporate IT: Insights into how the share-economy will play out?

When thinking about disruption we often attempt to look to the past for how it will affect our future. However, we can also explore how disruptive tech is impacting our world today.

Consider a recent WSJ article that gave us insights in to how cloud and mobile is disrupting "classic tech":

• EMC, Intel and IBM are being disrupted. The move to cloud and mobile is impacting the ability of these companies to meet earnings expectations. For example, EMC's sales of storage products declined by 10%, while Intel has seen an overall decline in PC Market sales.
• Companies are slashing their workforce in response to such trends. The shift to cloud & mobile is resulting has resulted in the hemorrhaging of 12,000 jobs, or 11% of the workforce, at Intel alone. 
• Overall decline in revenues/profits despite strategic shifts in product mix. IBM's cloud computing business grew 34%, while Intel's data centre business, serving cloud providers grew 9% (for more on this "corporatification trend" see here). However, IBM's total revenue fell almost 5%. Intel had a tough time keeping up with rivals like ARM who posted revenue gains of 22%. This compares to 7% of which the WSJ attributes a chunk of that rise to an acquisition Intel made. 

Reflecting on these trends, is it fair to think of cloud as the original use case for the share economy? 

When we think of Uber, Airbnb, etc., we see how users can use these platforms to monetize their excess or underutilized asset by renting it to others.

However, isn't that also the story of the cloud?

Amazon, Intuit, and other cloud computing companies decided to "share" their excess computing capacity to others. In a sense, we are talking about servers instead of houses, but the concept is really the same.

And this goes to my initial point. We are living through the disruptive impacts of cloud and mobile on legacy-tech and we can quantify, analyze and understand its impact.

Given this premise what does this above tell us about the share-economy?

Firstly, better utilization of assets leads to the sale of less assets.  This should be expected as there is a more efficient use of assets leads to  (servers, car, house, etc) less market size as demand remains constant. As I had noted in the post a few months ago, Uberization of the taxicab industry would ultimately lead to a fleet of cars that are owned by a company like Google - leading to a net reduction of cars used by society. GM probably understands this concept as they have invested $500 million into Lyft; a competitor to Uber.

Secondly, it illustrates how the share-economy is subject to concentration of wealth: the cloud computing landscape is dominated by large players, including Amazon, Google and Microsoft. As noted by Douglas Rushkoff in the following video, since these innovations emerge out of the "operating system of capitalism", they inevitably result in the formation of a handful of platforms dominate the industry and capture the lion's share of the profits.


(Also check out his book that discusses this in more depth)

Thirdly, its difficult for behemoths to adjust to these types of shifts. Despite Intel and IBM investing in the disruptive technologies, it's hard for them to adjust to the dynamics of the new economy. This illustrates how much more challenging it will be for those disrupted by such platforms to re-tool and compete in the landscape.

For accountants and auditors, one such platform to watch out for is Gigwalk. As per their website, their value proposition is that by leveraging the 350,000 “professional services” workers (see graphic below) the manufacturer or other upstream supplier can get visibility into the actual retail outlet. For example, Whirlpool wanted to “Audit the presence of its Swash product on showroom floors, communicating back to corporate compliance gaps in real-time”.

And the "starter Store Audit Package" begins at $10.

Creative destruction is not inevitable, but we should learn from the lessons of these tech giants and plan prudently to meet such challenges head on.

Artificial Intelligence: The new "space race" for the tech-giants?

When IBM's Watson defeated Ken Jennings and Brad Ritter on Jeopardy!, it was a shock. As Ken Jennings describes in this Ted Talk, he had no idea that a computer could possibly defeat him at Jeopardy! On this Ted Talk, Ken Jennings describes how he never thought that a computer could beat him:



And he's right.

How can a computer possibly understand that "feel can smell" and a "nose can run"? 

But on February 16th 2011, IBM's Watson did precisely that: it was able to defeat the two reigning human champions Ken Jennings and Brad Rutter. And with that IBM ignited the space race for artificial intelligence.

Although people may point to the wide array of personal digital assistants from Apple (Siri), Microsoft (Cortana) or Google Now as the true birth of the AI space race. However, these application are limited to the use of the personal arena. Anyone who used things like Google Now - which can link your calendar to traffic patterns and tell you if you'll be late for appointment - can tell how amazing it is to how have a digital assistant work behind the scenes to keep your day on track. That, however, is limited to the consumer realm. Where AI gets real interesting is the B2B realm: Watson has made some strides in automating the FAQ process. However, it's real promise has been demonstrated in the cancer treatment realm, where it enables doctors to "race with the machine" combining the millions of pages of medical journals and articles to determine the best cancer treatment for patients.  Watson is available in a cloud offering to developers who submit applications.

But IBM is not alone and so the AI Space Race is on!

As for the other vendors, see the following:

• Google's prediction API
• Amazon Machine Learning
• Microsoft's Azure Machine Learning

However, the one that I am really waiting to hear about is coming from the makers of the Siri, Viv.ai. They are hoping to build AI as a service, similar to Bluetooth, that will be embedded in all hardware. I will leave you with the following quote from the Wired article that discusses the possibilities of Viv.ai:

"Viv...generat[es] its own code on the fly, no programmers required. Take a complicated command like “Give me a flight to Dallas with a seat that Shaq could fit in.” Viv will parse the sentence and then it will perform its best trick: automatically generating a quick, efficient program to link third-party sources of information together—say, Kayak, SeatGuru, and the NBA media guide—so it can identify available flights with lots of legroom. And it can do all of this in a fraction of a second."

Time for Windows 10? I can't wait!

I have been overly optimistic about Windows in the past, but here me out!

Boy Genius published a post on the future Windows 10 that they are releasing next year. (Note Microsoft decided to skip Windows 9 altogether).

And does it looks good. As can be seen in this video, it will feature Cortana who is Microsoft's personal digital assistant that incorporates voice search, voice commands (i.e. you can get Cortana to set-up an appointment with you) and machine learning (i.e. it learns from your interactions.
Impressed?

It speaks to the overall move towards using natural language processing (NLP) and elements of Artificial Intelligence. Apple was arguably first to the scene with its Siri application. However, IBM's Watson is also a clearer example of where this technology is heading. Gartner refers to these types of technologies as "smart machines", which they claim has the following implications:

"Most business and thought leaders underestimate the potential of smart machines to take over millions of middle-class jobs in the coming decades," said Kenneth Brant, research director at Gartner. "Job destruction will happen at a faster pace, with machine-driven job elimination overwhelming the market's ability to create valuable new ones."

Will Cortana take your job? Well, let's just enjoy the possibility that Microsoft may build in some real cool NLP technology into your every-day computer and worry about that one a future post!

Big Data, Facial Recognition and Privacy

In October of 2012, the FTC released these guidelines on dealing with the privacy issues of facial recognition. The publication begins with a scene from the movie Minority Report where the protagonist is offered a product based on facial recognition software. When the movie came out, this future seemed decades away. However, now it just seems around the corner. And this is primarily due to advances in data mining capabilities brought to you by the cloud and more specifically big data. One of key characteristics of Big Data (see here for IBM's definition) is being able to include data that is not just massive or fast moving, but also goes beyond the simple world of flat file - meaning it includes images.

And here comes the interesting part about privacy and big data. Think about the following scenario: you walk  into a store, the security cameras take your picture, it's uploaded to Google Images (see below for how this works), a second search is done to find out what you've posted publicly on the various social sites and then you are approached by a sales associate who has all this information about you and then can tailor its offerings to you.

Privacy regulations requires that users give consent before their information is collected. So could this simply be circumvented by posting a sign that states "By entering this premises, you consent to the store collecting your image and using that information to tailor offerings, services and the like to your online profile"?

I think so.

As I posted last week, TJX was effectively able to stop the Canadian privacy watchdog, by agreeing to encrypt the personal information it should not collect. So it does not take much for companies to persuade users (or the privacy regulators) to get what they want from their customers/users. At the end of they day, companies simply have to repeat the Capitalist mantra: "You are free to go somewhere else if you don't like our policies". Never mind that all the companies have the same policies, which means you don't have any choice except to wear a mask when you enter the store. And that is not advisable because it may be result in being billy clubbed or tazered by security guards! Or to be a little less dramatic, the store will simply offer to refuse you service if you wear a mask.