Generative AI-based Conversational AI is a cutthroat business. Neither Microsoft Bing nor Google Search will be able to make higher profit margin out of it due to the sheer cost for training model, retraining model, human maintenance, data labelling, data sourcing, data cleansing, data de-biasing and ongoing human intervention when AI ChatBots utter offensive, biased or misleading answers. Besides, the ongoing huge expenditure spent in energy consumption (electricity bill), data centers (application servers, data storage), regulatory compliance and legal risk mitigation to sustain the AI ChatBot business is also larger than the traditional index-based search engine business. Since the ChatBot has offered broader context and much detailed explanation for every query or statement that users input, users no longer have motivation to click on the online advertisements that are being listed along with the ChatBot answers.
最近，ChatGPT掀起了基于生成式人工智能技术的聊天机器人风潮。在微软宣布其必应（Microsoft Bing）搜索引擎将逐步引进ChatGPT来增强用户的使用体验后，谷歌也仓促宣布目前正积极研发由聊天机器人Bard支援的最新搜索技术，并将在最短的时间内推出市场。
然而，由生成式人工智能聊天机器人支援的搜索引擎业务，却可能是割喉的生意战。不论是微软的必应搜索，还是谷歌的搜索，在推出聊天机器人来增强各自的搜索功能后，或都无法取得更可观的利润回报。这是因为应用聊天机器人来支援搜索功能的总体营运开销，将会比传统上利用索引条目来回应用户关键字搜索的业务支出来得高。这些额外的成本涵盖了培训模型、再培训模型、人工维护、数据标记、建模数据来源、数据清理、数据去偏差，还有防止聊天机器人导出冒犯性、充满偏见或误导性答案所需的持续人工干预与系统调整。
此外，聊天机器人业务更耗能（耗电）、更耗数据存储资源（应用服务器、数据存储硬件）、受制于更严苛的法规监管，以及面临对话内容涉及侵权或冒犯用户以至触法等挑战，也是代价不菲的开销。
最后，由于聊天机器人已经对为每个用户输入的文字查询，提供了更宏观的背景陈述和更详细的针对性解答，用户已经不再需要点击罗列在搜索答案页面上的在线广告。
参考资料/Reference
[2] https://www.macworld.com/article/1511542/macalope-google-microsoft-apple-ai-search.html
[3] https://www.businesslend.com/technology/when-it-comes-to-ai-maybe-apple-is-smarter-than-it-looks
Posts
0 Comments