ChatGPT disrupts Asian property industry
text size

ChatGPT disrupts Asian property industry

AI looks set to absorb some drudge work, freeing agents to focus on building relationships and providing valuable counsel, experts say.

Photo: South China Morning Post
Photo: South China Morning Post

HONG KONG: In a city of the future, a citizen looking to buy a home will simply explain their requirements to an artificial intelligence (AI)-based assistant, which will orchestrate the entire selection and buying process without a human property agent - or the commissions such agents command.

Well, maybe.

While real estate agents - a small army that in Hong Kong numbers more than 40,000 - may be wondering whether their jobs are in danger given the rise of ChatGPT and other AI services, the above vision remains science fiction.

As the real-world capabilities and limitations of ChatGPT and its ilk come into clearer focus, worries about AIs replacing millions of workers overnight appear overblown, at least for now. Yet across sectors, an age of AI experimentation has begun, with companies investing to understand the technology's potential advantages and pitfalls, as well as how it might impact their way of doing business. And property is no exception.

For example, Marcus Chu, Singapore and Asia-Pacific CEO of Singapore-based property agency ERA, sees AI as a golden opportunity to differentiate his company and his agents from the competition.

"When ChatGPT was launched globally in late 2022, which was the talk of the town in the tech world, we immediately seized the opportunity to customise ChatGPT into our app, known as Sales+, for our agents," said Chu, who is also CEO of APAC Realty, the parent of ERA Asia-Pacific.

APAC Realty holds the exclusive ERA regional master franchise rights for 17 countries and territories in Asia-Pacific, granted by New York-listed property services group Anywhere Real Estate, formerly known as Realogy Holdings Corp.

The agency has committed S$5.2 million (US$3.9 million) to technology tools, including ChatGPT. With the AI-enabled platform, ERA's agents can quickly and easily generate property listings, emails, and blog posts to a desired length and tone using automated guidance.

"We see great improvements in productivity levels, which means better client service," Chu said. "I believe ChatGPT has added significant value to our agents and that AI can benefit the real estate industry."

Aspects of buying or selling property still benefit from the human touch and experience of an expert, but there is still room for AI to take part in the process, according to industry leaders. (Photo: South China Morning Post)

ChatGPT is an AI chatbot built on a type of large language model (LLM) developed by US start-up OpenAI and released in November 2022. An LLM is a deep-learning algorithm that can recognise, summarise, translate, predict and generate text and other content based on knowledge gained from massive data sets. As such, it can answer questions and assist with tasks such as composing written material and even writing software code.

Chu noted that most agents, although they can talk very well, are not "natural born writers". Yet their jobs demand they spend many hours preparing written materials.

The Sales+ app can generate event invitations, personalised festive greetings, tenancy reminders and agent biographies. All content can also be customised and translated into the language used in the country where the agent is operating, according to Chu.

"The application of ChatGPT may lead to a reduction in the number of low-skilled jobs in the [real estate] industry, but it can also enable practitioners to focus on higher-value creative activities to drive innovation and improve efficiency," said Ryan Li, partner of management consulting at KPMG China.

"Standardised and repetitive work, such as property description and display, answering frequently asked questions, and market analysis, is very likely to be replaced by intelligent robot tools such as ChatGPT."

Tasks such as price negotiation, research into client demands and service customisation require more interaction, critical thinking and decision-making, Li added.

ERA Singapore has more than 8,700 agents and, as of the end of April 2023, the Sales+ app has been downloaded almost 14,000 times across multiple electronic devices. The ChatGPT feature in the app has been accessed more than 66,000 times since its March launch by more than 3,300 unique users, the company said.

ERA Singapore is also the largest international real estate agency in Singapore, operating across 11 countries in Asia-Pacific, with a combined workforce of more than 23,000 agents. Plans are already under way to roll out the ChatGPT features in the app to all ERA APAC offices in June.

With more time available, agents can spend more time with customers, Chu said. "They can grow their network and spend more time strategising on how to improve their sales conversion."

Centaline Group, which runs Hong Kong's biggest property-agency network, is also considering using ChatGPT, according to founder Shih Wing-ching.

"Secondary school graduates can be property agents," Shih said. "Since it is not an industry with very much an income guarantee, the level of knowledge among property agents is not high."

Artificial intelligence could help agents read and process legal documents and provide richer as well as more accurate answers, Shih said, noting that the business requires specialised knowledge of many ordinances and special situations.

"There are many facets," he said. "Some are legal. Some are financial. Using different people to answer different questions takes a considerable cost."

Shih emphasised that Centaline has not yet implemented AI. When it does, the agency would aim "not to replace, but to help" agents in their work and "enhance the accuracy of service", providing a broader scope of service at lower cost, he said.

Midland Realty is adopting ChatGPT for analysis of market statistics and client data to predict trends and provide personal recommendations; automation of business procedures; writing letters; analysing sales brochures; preparing sales proposals; video production; providing question-answering chatbots. The agency has organised 25 AI courses with over 2,000 attendees as of June 2.

Chinachem Group, a major Hong Kong developer, plans to launch two AI-enabled applications: "A corporate version of ChatGPT to help our colleagues work efficiently and free up time for creativity, and an AI image generator to streamline the process of generating required building-design images and reducing repetitive work," a spokesperson said.

In mainland China, where ChatGPT is not available, property platform Beike, owned by KE Holdings, has developed Xiaobei AI-Assistant, an intelligent online tool specifically for agents.

Before agents go out into the field, the tool offers online simulations that train agents in showing properties. In the field, it provides property recommendations and conversation tips. After an assignment, it evaluates agent performance and provides tailored training. The tool provided 136 million help suggestions and gave 73,325 reports on agent performance in 2022.

The benefits that AI may bring to the real estate sector look promising, but experts warn about the limitations and risks of the technology.

Putting in place measures to monitor and manage legal and ethical risks is critical in the design and implementation stages, according to international law firm DLA Piper.

For example, China's recent draft measures on regulating generative AI will be relevant to companies that use AI to provide customer services or generate valuation reports or market analytics.

Companies that feed customer data into AI systems must also comply with data-privacy and cybersecurity rules.

Additionally, proper measures must be taken to avoid bias and discrimination in AI systems, such as those that might occur in matching sellers and buyers, to reduce inaccuracy and mitigate against breaching anti-discrimination laws.

"The underlying algorithm of artificial intelligence must include screening and statistics based on big data," said Weng Guanxing, head of Shanghai-headquartered law firm Wintell & Co's Lingang Office. "Artificial intelligence can predict future market trends by analysing a large amount of historical data, and then guide the decision-making of real estate companies."

However, legal frameworks may make it difficult for companies to gather the large data sets AIs require for accuracy, Weng added. For example, with China's Personal Information Protection Law and Data Security Law, which went into force in 2021, the legal risks and costs of collecting personal data are increasing.

"While ChatGPT can replace some of the work and save agents time, AI technology has a long way to go before it can attempt to replace agents," said Laura O'Connor, chief operating officer of the US-based brokerage JPAR Real Estate's Affiliated Network.

The logo of OpenAI is displayed near a response by its AI chatbot ChatGPT on its website, in this illustration picture taken Feb 9, 2023. (Photo: Reuters)

The company is also using ChatGPT in ways similar to its Hong Kong peers.

"Our focus is on using emerging technologies to ensure that agents give reliable and prompt feedback to their customers," she said. "By reducing time spent on more rote tasks, it frees up time to focus more on homebuyers and their needs."

Despite continued attempts to digitalise the traditional real estate industry, the process of checking and signing real estate sales contracts cannot be made entirely paperless. And even though agents are technically not required in property deals, they bring value by smoothing the process for both parties.

Meanwhile, AI is finding use in property search. Two US real estate brokerage groups, Zillow and Redfin, separately launched their own ChatGPT plug-ins last month. These allow users to filter property listings in a conversational way, such as "show me homes that are in the US$475,000 price range".

Yet public perception of what AI can do seems to be far ahead of reality. For example, Hong Kong home seeker Kary Yang was disappointed after testing out ChatGPT in her search.

Yang, who works as an immigration and cryptocurrency exchange license consultant, subscribed to ChatGPT's pro version and asked it to identify a HK$5 million flat (US$638,000) in Hong Kong's Southern district.

"It is possible to find a HK$5 million flat in Hong Kong's Southern district," the bot responded. "However, the size and location of the flat may vary." It then cited old news articles about the prices of Hong Kong's homes, followed by a conclusion that merely reiterated the first sentence.

"ChatGPT only gave some general advice but did not give any specific property recommendations," Yang said.

This is not surprising given that the public versions of ChatGPT only have access to information up to 2021 - using Google or a property site would have been far more fruitful.

The experience led Yang to the correct conclusion about AI's use in the property field: "It cannot replace property agents."

Only time will tell whether that statement remains true.

Do you like the content of this article?