Google vs. Microsoft: The good, the bad, and the ugly of the AI ​​arms race

Over the past few weeks, the competition between Microsoft and Google for large-scale language models has been escalating. The two tech giants are locked in a tug-of-war over how people will search for information in the future. Or, more accurately, Google is trying to protect its search business from Microsoft and OpenAI’s massive language models.

Since OpenAI released ChatGPT last November, there has been a lot of speculation about the killer application of this large language model. One of the recurring topics is whether ChatGPT and other large language models will make Google search obsolete. Some industry experts believe that chatbots like ChatGPT will replace Google search, which could pose a huge threat to Google because it generates most of its revenue from search business.

Overall, the competition between Google and Microsoft may be positive for the development of the search engine market. But it also has a dark side, particularly how it will impact smaller AI developers and the future of AI research. And the uglier side of Google and Microsoft's hasty actions could have broader consequences for Internet users.

Benefits: AI brings innovation to search

Google has always been the undisputed leader in the search engine market, with more than 90% share and a huge User group and advertising network, it is also the developer of two of the world's most popular browsers (Chrome and Safari), as well as the default search engine for Android and iOS systems.

This dominant position makes Google enjoy the vast majority of the market share without any impulse to innovate. Sure, Google Search has seen some improvements over the past few years, including better answer capabilities and deep learning enhancements, but the core of its search experience hasn't changed. Many users have complained that the search experience has gotten worse over the years, with more ads.

Microsoft now plans to offer a new search experience on its Microsoft Edge browser, allowing users to use the classic search model as well as a conversational interface powered by ChatGPT.

This move by Microsoft is very likely to make a profit, and even if it fails, it will not suffer much loss, because its own share of the search engine market is negligible and there is a lot of room to try new products. and business model.

On the other hand, Google may lose everything, because the search business is Google's largest source of revenue, and losing even a small part of its market share will have dire consequences.

Currently, Google’s response to Microsoft’s Bing Chat GPT project was a hasty demo of its own Bard, which unfortunately misrepresented the facts and resulted in $100 billion of the company’s market value evaporating.

Google must now innovate and reinvent itself or see other tech companies cannibalize its search engine business. The fierce market competition is not over, and Google still has a lot of money to solve this problem.

Disadvantage: AI research will be concentrated in technology giants

As competition intensifies, technology giants try to defeat their opponents in various ways. One of them is to launch new technologies and add features to existing AI products. But another shortcut to innovation is to cooperate with some promising startups and research labs, maybe even acquire them, or provide them with more resources and protection.

People can see the clues from Microsoft’s recent multi-billion dollar partnership with OpenAI, and Google’s recent $300 million investment in Anthropic, an AI research laboratory founded by former OpenAI scientists.

In return for the investment, Microsoft has obtained an exclusive license to OpenAI technology. Microsoft Azure has become the exclusive cloud computing provider for OpenAI research. Google Cloud Platform has also become an Anthropic provider.

As the AI ​​competition between Google and Microsoft intensifies, this cycle will repeat itself. The two tech giants are trying to outspend each other to attract more AI startups to their ranks. They also have many options, such as Cohere AI, Stability AI, Midtravel, You.com, Perplexity A, Copy.Ai, and Hugging Face.

While these startups and research labs, most of which are not yet profitable, will benefit from the infusion of cash and subsidies, they will also be caught up in the battle between Google and Microsoft. Their research will gradually lose freedom and flexibility and become subject to the short-term interests of financial backers. This means they will research more on technologies that can be monetized quickly and less on unexplored or non-profit technologies.

People have seen this with OpenAI. While large language models and generative AI are fascinating and offer many capabilities and uses, they are not the only promising areas of research. In order to focus more on the kind of technology that Microsoft will benefit from, there are signs that OpenAI has lost interest in other areas such as robotics and game AI.

The Ugly: Bad AI Will Flood the Internet

Currently, people are keen to seize market share in the field of generative AI, which is different from Facebook's early "move fast and break the ice" era. Because these two tech giants are very large companies with billions of users, any changes to their technology and products will have an immediate impact on the global market.

But these two companies are increasingly eager to launch new AI features, and the more hasty decisions are made, the more devastating consequences they will have.

To some extent, this reminds people of the enthusiastic pursuit of Internet of Things technology by enterprises about 10 years ago. Manufacturers have rushed to jump on the smart device bandwagon, launching their own IoT solutions, but these solutions are often riddled with security holes, resulting in billions of unsecured devices connected to the Internet, many of which have no way Patch vulnerabilities.

They are later used by cyber attackers in large-scale distributed denial of service (DDoS) attacks, espionage and other malicious activities. It took a number of disruptive incidents for device manufacturers to seriously consider developing security standards for the Internet of Things and the Industrial Internet of Things (IIoT).

Meanwhile, people are still exploring the legal, ethical, and social implications of large language models. The generation of factual errors by large language models like ChatGPT is a problem that has not yet been fully solved. For example, websites can use them to generate articles that are grammatically correct but factually incorrect, which ultimately dilutes the web and search results, generates a lot of bad content, and makes the user experience even worse.

There are also concerns about students using ChatGPT to complete assignments or cheat on exams. While this is a minor issue, it suggests that many social structures, including education systems, now need to adjust themselves to this new reality. These changes may not happen as quickly as tech companies rolling out products powered by large language models.

People’s biggest concern is data security and privacy. To make large language models work for all organizations, tech companies have created tools that allow users to fine-tune models using proprietary data without data engineering skills. However, training and fine-tuning machine learning models in the wrong way can lead to models leaking sensitive data. As the integration of large language models into production tools becomes more common, this can turn into a nightmare.

In short, research such as large-scale language models and generative AI can bring huge benefits to mankind, but if technology companies fall into an immeasurable market war, it may bring bad consequences. Unfortunately, this may be happening!

The above is the detailed content of Google vs. Microsoft: The good, the bad, and the ugly of the AI ​​arms race. For more information, please follow other related articles on the PHP Chinese website!