The march of technological advancements throughout human history has witnessed several revolutions, with the introduction of steam engines, electricity, and the internet being some notable examples. But in recent years, we’ve embarked on the dawn of another significant shift: the age of Artificial Intelligence (AI). 

This new wave of AI, with its promise and complexities, was one of the discussions at the CRE.Converge 2023, a conference facilitated by the National Association for Industrial and Office Parks (NAIOP). NAIOP, recognized as the leading organization for developers, owners, and professionals in office, industrial, retail, and mixed-use real estate, supports its members through education, networking, and legislative advocacy, all while promoting responsible, community-beneficial development. 

It was during this conference that Smart Capital Center took center stage, having been chosen by NAIOP to present the transformative impact of AI technology such as ChatGPT at the CRE.Talks in Seattle. The focus was on the transformative power of ChatGPT and related technology, particularly in the investment and asset management processes. 

This event, marked by standing-room attendance, underscored the growing interest in AI today. 

GPT-4, or the Generative Pre-trained Transformer 4, is a state-of-the-art machine learning model designed by OpenAI. As the latest in a series of models, GPT-4 is the culmination of years of research and development, building on the successes and learnings of its predecessors. It is built to understand and generate human-like text based on the vast amounts of data it has been trained on. Its primary capability lies in text generation, but its versatility allows it to tackle a myriad of tasks, from simple text completions to more complex tasks such as language translation, question-answering, and even some basic forms of reasoning. 

The infiltration of AI in business isn’t a new narrative. However, the launch of ChatGPT, just about 11 months ago, has given a glimpse into the future possibilities of AI. We’re now entering a phase where businesses and individuals are grappling with the potential of Large Language Models (LLMs) such as GPT-4. 

Even the creators, the scientists, and developers behind these monumental AI models, can’t fully predict the ripple effects their inventions will cause in different industries. 

Current understanding, therefore, is largely empirical. Researchers, scientists, and curious individuals are pioneering ways to harness the capabilities of LLMs, continually testing, iterating, and learning. They’re navigating the landscape with a combination of structured research and exploratory forays. 

The true mastery over the new wave of AI, such as any tool or technology, will come from consistent application, experimentation, and a willingness to evolve and adapt. The age of AI, driven by models such as GPT-4, is just beginning, and the horizon holds promises and challenges we’re just beginning to fathom–especially for the real estate industry. 

The Power of AI in Enhancing Individual Productivity 

Laura Krashakova, CEO of Smart Capital Center, opened the discussion by highlighting that the rise of artificial intelligence has ushered in a new era where machines and humans are no longer seen as competing forces but synergistic collaborators. At the forefront of this transformational shift is the evidence pointing to substantial benefits that AI brings, particularly at the individual employee level. 

According to Laura “One of the most significant advantages, as gleaned from early observations and controlled studies, is the profound impact on individual productivity. With the integration of AI tools into routine tasks, we are witnessing potential time savings ranging from a significant 20% to a whopping 70%.” 

These tools have not only made work faster but also improved the quality significantly, leading to tasks being done more quickly and with better precision and excellence. This means that not only are tasks being completed faster, but they are also executed with greater precision and excellence. 

A study that underscores these observations was conducted by a team of social scientists in collaboration with Boston Consulting Group (BCG). This study wasn’t just any ordinary research venture but was designed to mimic the future of professional work in an era increasingly influenced by AI. 

In this experiment, 18 diverse tasks, representative of the kind of high-quality work typical at elite consulting firms such as BCG, were chosen. The results were indisputable. 

Consultants equipped with the power of ChatGPT-4 significantly outperformed their peers who didn’t have the AI tool at their disposal. This outperformance wasn’t limited to one metric or a single dimension but was universal across all performance indicators employed in the study. 

The study discovered that consultants using ChatGPT-4 showcased remarkable improvements in their performance metrics. They accomplished 12% more tasks, completed their tasks 25% faster, and what’s even more commendable is that the quality of their output was elevated by 40% compared to those working without the assistance of ChatGPT-4. 

What’s crucial to highlight here is that these remarkable results were achieved even before the most recent advancements to GPT-4 were introduced. This means that with the introduction of advanced data analytics modes, plugins, and up-to-date web searches, the potential benefits are likely to be even more significant. 

AI, particularly models such as GPT-4, are not just tools for efficiency; they signify a future where human intelligence, augmented with artificial intelligence, can achieve unparalleled feats, setting new benchmarks for quality and productivity. 

Beyond ChatGPT: Exploring the Wider AI Universe 

Laura also observed that “ChatGPT is the best among the options available.” 

Chat GPT in the AI Landscape: 

• Diversity of AI Models and Apps: While there’s an abundance of AI models, only a few truly dominate the scene. ChatGPT, currently, is the best. 
• Notable LLMs to Consider: Apart from ChatGPT, there’s Claude 2, a product developed by Anthropic with funding from Amazon. Then, there’s Bard by Google. Although Bard lags behind in comparison to ChatGPT and Claude 2, the anticipated release of Google, Gemini, is rumored to leapfrog the current capabilities of ChatGPT. 
• The Broad Spectrum of LLMs: Companies such as Facebook, Amazon, and numerous Chinese tech giants have ventured into developing their own LLMs. 

Public vs. Private LLMs: 

• Public LLMs: ChatGPT is a notable example. It sources data from global users and other openly accessible information. There’s no provision for hosting it on private servers, yet. 
• Private LLMs: Tailored with specific datasets and can be hosted privately. While they ensure data security, their performance hasn’t caught up with the likes of ChatGPT. 
• Open Source LLMs: Businesses wanting to host LLMs on their servers with data control can opt for either open-sourced or non-open-sourced enterprise-focused LLMs. 

At Smart Capital Center, initial prototyping is executed on ChatGPT-4 due to its advanced capabilities. But for production, different LLMs can be deployed based on specific use cases. Despite developing its own private LLM, the company continues to employ ChatGPT to tap into the expansive possibilities of AI for future initiatives. 

With the fast progression of ChatGPT technology, adopting a wait-and-see strategy could be advantageous. The features Smart Capital Center introduced in June for data and document analysis through ChatGPT-4 have since become more sophisticated and widely adopted. 

Sometimes, waiting is advantageous. Simply delaying a few months can yield better results than developing an earlier version of the technology. 

From Summaries to Strategy: Applications of LLMs in Investment & Financing Workflows 

1. Summarization and analysis of written materials, including web pages, PDFs, scanned, and other documents. 

In today’s information-driven world, the ability to distill vast amounts of textual data into comprehensible insights is invaluable. With the evolution of AI, particularly models such as ChatGPT and other LLMs, the horizon of summarization has vastly expanded. 

While human analysis remains irreplaceable, it often becomes a time-consuming task to dissect massive documents such as Info Memos, Market Overviews, or Due Diligence Materials. AI-driven models promise efficiency in this aspect. With their ability to scan and process information at unparalleled speeds, these tools can condense vast sections of text into summaries. 

The efficacy of an AI summary largely hinges on how the user communicates with the AI – this is where ‘prompt engineering’ comes into play. Becoming adept at framing prompts is essential to harness the full potential of AI models. These prompts guide the AI on what to focus on, ensuring the output is aligned with the user’s expectations. 

It’s fascinating to observe how prompt engineering has evolved. Initially, a simple question or directive would suffice. But as the complexity of tasks grew, so did the sophistication of prompts. Today, it’s not uncommon to find prompts that span one or two pages, particularly for intricate analyses. 

Research teams globally have dedicated themselves to mastering this art. Through rigorous experimentation and testing, they’ve identified techniques that optimize the performance of AI. Their findings, often shared in published papers, are a treasure trove for those keen on utilizing AI to its full potential. 

A noteworthy advancement in this domain is the ‘Chain of Density’ technique.  

Traditional summarization methods, when executed by AI, might inadvertently miss out on vital information. The Chain of Density addresses this shortcoming. It requires the AI to execute multiple rounds of summarization, incrementally condensing the information while ensuring critical details aren’t omitted. 

Interestingly, this technique has seen further enhancements. A variation called ‘Chain of Density with user editing’ has emerged, allowing users to intervene during the summarization process of AI. This ensures that the user’s expertise complements the efficiency of AI, resulting in a summary that is both concise and comprehensive. 

At Smart Capital Center, we are already using the power of AI. We have integrated ChatGPT into our systems, empowering users to analyze complex CRE documents—such as loan documents, insurance documents, disclosures, and lease agreements—with impressive accuracy and speed. 

2. Analysis and visualization of numerical and other data from tables, databases, or API data. 

As businesses continue to leverage vast amounts of data for insights and decision-making, the importance of effectively analyzing and visualizing this data cannot be understated. One area where this comes into play is numerical data obtained from tables, databases, and API data. 

While ChatGPT boasts impressive capabilities, it’s worth noting that it initially faced challenges in direct data analysis. Early versions of GPT-4, in particular, had noticeable limitations when it came to computations. 

Recognizing this gap, the developers incorporated the Advanced Data Analysis submodel into the latest GPT-4 iterations. This new model boasts a code interpreter that seamlessly crafts Python code. After running this code on the given data, it then translates the results back to the user. 

To further bolster the computational capacities of GPT-4, plugins such as Wolfram Alpha are often used for intricate calculations. In fact, GPT-4 introduced a dedicated submodel named “Plugins” that facilitates connections to external applications. For those familiar with Wolfram Alpha, it can even be integrated directly within GPT-4. 

After data analysis, presenting the findings effectively is equally crucial. GPT-4 partners with the statistical application, MATLAB, to produce charts. Admittedly, these visual representations aren’t the most aesthetic. However, users can leverage third-party libraries, such as Diagrams, to enhance the visual appeal of data presentations. 

While ChatGPT-4 possesses extensive capabilities, it is not optimized for interpreting unstructured data formats, such as scanned documents. Uploading a scanned financial statement for direct analysis would not yield accurate results, as the model requires data in a more structured form for effective processing. 

However, there is a solution. Smart Capital Center users can use the platform’s preprocessing capabilities to convert financial documents, rent rolls, operating statements, projections, or even ARGUS files into structured and standardized data. Once preprocessed by Smart Capital Center, the data can then be effectively analyzed by GPT-4. 

Another limitation is the potential for inaccuracies in outputs of ChatGPT. This necessitates a review process, especially when precision is paramount. Fortunately, the system is evolving rapidly. In instances where discrepancies arise, such as misinterpreting subtotals, Advanced Data Analytics mode of ChatGPT-4 can often provide context and recommend solutions. It underscores the importance of a human-AI collaboration, where the user’s domain knowledge complements the computational power of AI. 

Smart Capital Center addresses this by offering accurate interactive analyses. By incorporating GPT-4, Smart Capital enhances these analyses, making them more dynamic and allowing users to create their own analytics in real time. This integration ensures that while GPT-4 provides computational power, Smart Capital Center ensures the accuracy and relevance of the output, tailored to the user’s needs.

3. Generation of materials, such as draft investment memos and other documents.  

Drafting an investment or credit memo has traditionally been a labor-intensive process rife with repetitive tasks. Enter ChatGPT. With its advanced capabilities, it can now generate initial drafts for select sections, pulling from both internal and external data sources. Whether it’s a market summary, property insights, or a SWOT analysis, GPT can set the foundation, complete with custom charts, simplifying the entire process. Starting with a preliminary draft is far more efficient than starting from scratch. 

However, to optimize its capabilities, the key is collaboration. Instead of solely relying on GPT for the first draft and then passing the baton to a human, a more iterative approach is recommended. By constantly engaging with GPT, refining inputs, and incorporating new data, the end product is far more comprehensive and accurate. The synergy is particularly potent when GPT is integrated with an underwriting platform, enabling seamless access to precise property metrics, deal specifics, and visual data such as charts and maps. 

 4. Analysis and key extraction of key terms from complex legal documents. 

Legal documentation offers another avenue where GPT shines. It can be employed to sift through complex loan agreements, leases, and similar legal texts, extracting essential terms and metrics. Consider the intricate covenants in loan agreements or terms in leases; while the initial foray of ChatGPT into this domain earlier this year left room for improvement, its current performance is notably enhanced. 

Beyond analysis, the power of GPT extends to the creation of legal documents, varying from formal agreements to routine emails and letters. Many professionals are already exploring these capabilities, with some adopting them as part of their regular workflows. 

Laura also added, “The interest in legal AI is evident from the staggering investment figures — over a hundred billion dollars directed towards startups in this niche. As someone perpetually on the hunt for the quintessential AI assistant, I’ve observed that while many offerings are promising, there’s still a journey ahead to perfection. However, the strides made are undeniable. Some platforms can already draft preliminary agreements, and it’s only a matter of time before the bulk of legal drafting becomes automated, thanks to the relentless evolution of AI.” 

ChatGPT & Data Security: Overcoming Corporate Hesitancy 

Large Language Models such as ChatGPT, when integrated into corporate workflows, have been met with hesitancy. Concerns about data security and potential breaches have led numerous corporations to enforce prohibitions against using such tools. However, as we delve deeper, it’s clear that these apprehensions, although valid, may be transient. 

There are several compelling reasons to believe that LLMs will eventually find their place in corporate systems, with data privacy concerns addressed: 

1. Local Deployment: Corporations can now deploy LLMs such as ChatGPT on their own servers. This means sensitive data need not exit the corporate firewall, effectively preventing any data leakage to external servers or third parties. 

2. Enhanced Privacy Features: Public LLMs have started integrating privacy-centric features. For instance, OpenAI and Microsoft have been vocal about their commitment to data privacy. They offer features where individual users can activate a “privacy mode,” ensuring that the data fed into the LLM isn’t retained or used for training the AI further. It’s an initiative that underscores the industry’s shift towards enhanced security and compliance. Along similar lines, there are now HIPAA-compliant versions of LLMs, a move that will certainly bolster trust among users. Claude by Anthropic serves as an example, having attained HIPAA compliance. 

3. Anonymization Techniques: LLMs are capable of processing data that has been anonymized. With Smart Capital’s data processing and deidentification features, users can utilize ChatGPT for extensive data analysis. 

In the larger picture, while initial reactions to any new technology might veer towards caution, it’s essential that companies transition from broad prohibitions to more nuanced policies. By pinpointing specific use-cases and developing targeted guidelines, corporations can harness the potential of LLMs without compromising on data security. The key lies in evolving from blanket bans to informed and adaptive strategies. 

Bridging the Skill Gap: How AI Levels the Professional Playing Field 

The transformative impact of AI on productivity, especially among professionals, has been significant but not uniform. In numerous studies, it was observed that AI especially benefited those traditionally deemed as ‘lower-performing’ by boosting their output quality and efficiency. 

Laura commented, “We’ve seen undeniable evidence that AI acts as a skills equalizer across various professional tasks. Whether it’s writing, idea generation, or analysis, individuals previously in the lower half of skill distribution have found themselves elevated to levels of competence, often surpassing the former averages, with the assistance of AI.” 

Daniel Schwarcz, an award-winning professor, and scholar, together with Jonathan H. Choi, a fellow professor at the University of Minnesota Law School, have found through their research that law students, when assisted by AI, can perform at levels comparable to or even surpassing those of their higher-ranked peers. This trend is mirrored in the customer service sector, where employees with lesser performance metrics have shown significant improvement with the support of AI. 

However, this uniformity brought by AI might be fleeting. Future AI models might offer top-performers even larger productivity boosts. Some individuals inherently excel at integrating AI into their workflows, yielding disproportionate benefits. 

Interestingly, optimal results were achieved when employees collaborated with AI in an iterative manner, refining outputs and incorporating AI suggestions progressively rather than offloading entire tasks to the machine. For these individuals who are good at AI, the technology isn’t just an aid but a game-changer, positioning them as the sought-after stars in the AI-driven era. 

Yet, while individuals harness AI to boost personal productivity, organizations as a whole might not reap the same advantages. The ‘secret cyborg’ phenomenon – where employees utilize AI advancements discreetly without notifying their superiors or peers – can be a double-edged sword. It underscores the pivotal role organizational culture plays in the integration and open adoption of AI. Companies fostering a cooperative environment will likely see a more transparent and holistic adoption of AI, as opposed to those with competitive cultures. 

As we venture deeper into the AI era, it becomes essential for organizations to foster an environment that encourages sharing, learning, and evolving, ensuring that the benefits of AI are widespread and not just limited to individual ‘secret cyborgs’. 

The Future of Work: Embracing the Transformative Power of AI 

Our professional lives encompass managing a series of complex responsibilities, not just executing automatable tasks. As AI enhances different aspects of these roles, it doesn’t suggest human redundancy but rather frees us. 

Smart Capital Center, for instance, removes much of the routine, tedious work, enabling real estate analysts to concentrate on more substantial tasks such as higher-level analysis, actual underwriting, and strategic decision-making. This shift allows professionals to engage in more significant, fulfilling work, utilizing AI to handle the more mundane aspects. 

Yet, as we enter this new age of AI, numerous questions loom large. High-earning professions seem more susceptible to AI advancements, prompting companies to ponder their future workforce dynamics. Will they opt for less-skilled individuals, boosting their capabilities through AI? Will they demand heightened productivity from their current roster?  

Perhaps the focus will shift towards nurturing machine-augmented super performers. Or, in the quest for profit maximization, will they contemplate trimming salaries or even downsizing? 

Deciphering these problems will be a collective endeavor shaped by corporate decisions, governmental policies, employee aspirations, stakeholder interests, and customer preferences. 

What remains undeniable is the evolving power of AI. AI-powered platforms such as Smart Capital Center can be leveraged today by investors to redefine the landscape of financial analysis and investment management. Through the use of advanced technology, these platforms provide in-depth investment analysis, offering faster and more accurate market and property financial assessments and helping in making more informed decisions. 

Optimized portfolio management becomes a reality with AI, as it dramatically speeds up the analysis of performance metrics and forecasting trends. It condenses vast arrays of data into actionable insights, turning what would be hours of analysis into minutes and driving significant cost and time savings. 

In the current financial climate, every investor’s toolkit now demands a grasp of AI and automation—and Smart Capital Center stands out as the solution of choice, equipping users with the tools necessary to thrive in this new era of CRE. 

Smart Capital envisions a future where the fast adoption of digital and AI technology will greatly accelerate the pace of transactions. It’s imperative for real estate investors to adopt AI, perhaps reluctantly, just to remain competitive in the marketplace… just to survive.  

The concern is not AI itself, but rather the risk of falling behind competitors who are quick to harness its power.