ByteDance vs. Baidu – AI bake-off.

A test of ByteDance’s AI.

• By moving into search, ByteDance is throwing down the gauntlet to Baidu where the contest will be decided by the quality of the algorithms that each company uses.
• My money is on Baidu but there is no doubt that ByteDance represents a threat to Baidu’s long-term search business.
• ByteDance is a new entrant into the Chinese digital ecosystem landscape and has come to prominence as a provider of curated news articles and short videos not unlike SnapChat or Instagram.
• It has been successful mainly by its ability to work out the kind of videos and news articles that its users like and then recommend suitable content.
• It is its success at doing this that has driven usage and the spread of its apps both in China and overseas.
• TikTok is pretty much the only Chinese Digital Life service that has gained any traction overseas further underlining that this is one area where the West copies the East.
• Typically, the Chinese come up with a service or an idea that works in China and then Facebook or Google will Westernise it and launch a similar version within their own ecosystems.
• This overseas success makes ByteDance an outlier indicating that its TikTok video service offers a good user experience and is backed up by good algorithms that keep the content fresh and relevant.
• This is an indication that the quality of its AI is better than most and worthy of consideration.
• ByteDance’s AI lab was created in 2016 by Wei-Ying Ma who came from Microsoft Research and it focuses on video, image and text analysis and developing recommendation engines.
• This is a good skill to have but it is quite a narrow focus compared to Baidu where the user can ask anything of the engine and often receive a relevant answer.
• It is this breadth and the ability to deal with almost any request that allows Baidu to standout.
• RFM research has indicated (see here) the best indicator of excellence in AI is not the amount of data but how long the AI company has been crunching that data.
• The search companies Google, Baidu and Yandex have been crunching data for over 20 years which is why they are currently the best at AI.
• ByteDance which began its efforts on 2016, is unlikely to overturn this advantage overnight but its narrow initial focus may allow it to gain some share in the domains where it operates (news and video).
• For example, ByteDance will be implementing this function in its apps meaning that the searches it carries out will be for videos, news items or related products.
• The aim is to earn a return through the sale of targeted advertisements based on the search terms or the user’s history.
• Consequently, I think that ByteDance has an opportunity to put a small dent in Baidu’s search business largely through the convenience searching from within the app but when it comes to general search, this has yet to become a major challenge.
• Hence, this is a small negative for Baidu but nothing that is likely to materially impact its search revenues for at least 24 months.
• Baidu has had a pretty rough 18 months, but I still rank it as the global No. 2 in AI making it one of the cheapest ways to invest in AI available.
• I am keeping my eyes open for the right time to come in.