In a recent MarketWatch article, I predicted that Alphabet’s GOOGL,
Google will be forced to position its digital chiefs and provide a worthy challenge to ChatGPT if it wants to remain the king of the search engines. That time has come.
Google subsidiary DeepMind is preparing to launch Sparrow, which is said to be a safe, “harmless” and reliable alternative to Microsoft-backed MSFT,
ChatGPT. Although the exact date has not been confirmed, DeepMind intends to test the private beta later this year.
“ AI has been learning from the internet, which is often apathetic, biased and inaccurate. “
I have been using ChatGPT since its existence, so I know both its shortcomings and its great potential, if one knows how to use it correctly.
Its shortcomings are the result of several problems, from inconsistencies in AI model training to dataset-based limitations. Simply put, AI has been learning from the internet, which is often biased, biased and inaccurate. ChatGPT’s limitations, for example, include occasionally providing incorrect information and dangerous instructions or biased content.
Sparrow can deal with most of these problems because it can draw on a very clean dataset, thanks to its direct access to Google’s most advanced semantic search engine and decades of research.
DeepMind will likely improve the dataset and use filters to fine-tune the model to provide more accurate information that comes with active backlinks, which the OpenAI model cannot currently do.
As a result, Sparrow is more likely to get stuck in what he can do. It makes sense since Google is using it to quickly fix a threat to its search engine empire rather than offering a variety of AI models.
In the first test, Sparrow was able to give an honest answer and back it up with evidence 78% of the time when he was asked a factual question. These performance results are impressive, especially early in the development cycle, and there is a good chance that Google’s AI model will perform very well when released.
However, the clash of AI giants is following something pretty bad – for Google at least. The company started developing Sparrow because it was afraid that people would end up relying on ChatGPT instead of its engine for their search queries.
Even if Google wins this battle, it will have to take the next step: integrate Sparrow into Google’s search engine/ad serving business. This will be an expensive and dangerous repair.
The Sparrow will have to work seamlessly with the current engine, maximizing its results and contributing to the company’s formidable marketing plan. Adding an additional AI model to a tightly coupled ecosystem and continuously training it on highly variable datasets (so that it becomes a useful piece of analytics rather than a loose cannon spewing out dated information and nonsense) will be a challenging and expensive task.
It’s a bullet Google will have to bite, as its competition doesn’t have a $209 billion a year ad empire to protect and can “move fast and break things” with fewer results.
Speaking of results, there may be a few on Google if things go wrong. If the search engine master integrates the model too early and starts to give misleading or biased answers at critical times, such as elections, or about divisive social and political topics, it may cause expensive problems for Google — those that may question its reputation and lead to various legal proceedings.
On the other hand, Microsoft’s subpar search engine, Bing, may finally get the boost it needs to be removed from obsolescence now that the company bought an exclusive license for the underlying technology behind GPT-3, the predecessor to ChatGPT, back in 2019.
In addition, unlike Google, Microsoft does not have to protect the current ad revenue stream (at $7.7 billion in 2020, there is a lot of room for improvement), and we can expect Microsoft to eagerly and aggressively integrate the model into its pipeline.
Although both AI models have their strengths and limitations, the success of Sparrow and ChatGPT will have a major impact on the future of search engines and online advertising. The battle between these AI giants will also impact data privacy, information accuracy and the role of AI in society.
More: People are crazy about ChatGPT. Here is a practical application for advertisers, programmers and journalists