Google's (NASDAQ:GOOG) (NASDAQ:GOOGL) recently released Allo instant messaging (IM) app has received mixed reviews at best. Despite the inclusion of powerful artificial intelligence (AI) features, Allo lacks many features that are taken for granted in IM apps. What was Google thinking? Google was thinking about the future, but may have lost track, temporarily, of the present.
Source: Ars Technica
Say Allo to AI
It was following the Google IO keynote in May that I first wrote about Google's new Allo and Duo (for video calling) apps in "Rise of the (Learning) Machines". Starting with Microsoft's (NASDAQ:MSFT) Build developer conference in March, AI was to become a key theme of all the major developer conferences, including Apple's (NASDAQ:AAPL).
Google appeared to have the best AI creds, having created AlphaGo which defeated Go champion Lee Sedol. Google's approach also appeared to be the most compelling. It would embed Google Assistant (GA) in a variety of apps, including Allo. Google Assistant would have search and voice recognition capabilities based on existing Google apps, but would also be capable of learning from the user's actions and queries.
Google predicted that this learning capability would improve speech recognition as well as allow the assistant to suggest useful replies in Allo. How well this works in practice seems to be a matter of opinion.
Ron Amadeo, writing for ArsTechnica, liked the powerful search capability that GA makes available within Allo. Brian X. Chen, writing for the NY Times, was not impressed by GA's attempts at suggesting message responses:
When I sent a picture to a friend of my cat sitting inside my car, Allo suggested this response to the friend: "What a cute car!" (Sorry, Allo, but my Prius is the opposite of cute.)
When I sent photos of my dog to the same friend, Allo's assistant correctly identified the breed, a Pembroke Welsh corgi. It suggested the reaction "Nice pembroke welsh corgi." Impressive, but if someone said that to me in real life, I would add that person to my list of suspected Cylons.
Or suspected Google Assistants. I don't think anyone really expects that much from the AIs that any of the three major personal computing platforms, Windows, iOS and Android, can offer at this point. Most of the criticism concerning Allo was for all the things that got left out.
It's a fairly long list. Allo doesn't support multiple user devices, but only works on one device at a time. It doesn't support SMS. It only works as a scaled-up smartphone app on a tablet. It uses the smartphone's contact list, rather than its own list. Amadeo was particularly annoyed that he couldn't even sign in using his Google account. All these things that Allo doesn't do are pretty much standard among its competition.
The India Thesis
The lack of so many standard IM features led Amadeo to propose an explanation in a separate article that Allo had been targeted at the Indian market, where most users have only a single mobile device. He points to the fact that Allo was rolled out in India first. He also wondered why, if Allo was intended for India, it was ever released globally.
To flip that around, the fact of the global release is probably the best counter argument to India thesis. Certainly, the impression one got at IO was that this was going to be part of a major global push. Google is trying to insert GA wherever it can.
I think it more likely that the other features fell on the floor because implementing GA just turned out to be more difficult and costly than expected. GA had the highest priority, so other things fell by the wayside. This may also be indicative of a more schedule-disciplined approach to development. Allo was expected at roughly the same time as Android 7.0, which came out at the beginning of the month.
Whether releasing Allo at this point was really a good idea, given all the missing features, time will tell. I'm pretty sure that Google understands that these features need to be included and will fold them into future updates. Then, Allo will truly be a global app and not just "for India."
The AI Conundrum
One of the issues that drew some criticism was the fact that end-to-end encryption, known as Incognito mode, was not on by default. Google could have called this something a little less sinister, like "Privacy mode," but I guess privacy is kind of a sensitive issue for Google.
The reason for the default appears to be that GA monitors the messages as they come into Google's servers, and it can't do this when they're encrypted. This tension between privacy and the need to enable machine learning is probably always going to exist to some degree. Google Assistant can't learn from your messages if it can't read them. Users just have to decide for themselves which is more important, privacy or convenience. I suspect most will choose convenience.
One possible solution is locally hosted AI. An AI resident in the user's smartphone would allow end-to-end encryption for both incoming and outgoing messages, while still being able to provide some assistance. It's not a perfect solution, but it does preserve privacy.
We're already seeing fairly competent AIs being implemented on smartphone processors for products such as Hover Camera, and Qualcomm (NASDAQ:QCOM) has released a set of APIs to implement AI functions on the Snapdragon 820. So this is definitely feasible, and we'll probably start to see more of this from companies such as Apple that want to preserve user privacy. Locally hosted AI could become a major driver in mobile processor performance for the next few years.
Despite the apparent stumble with Allo, I still think Google's on the right track. It represents a breakthrough in bringing AI even closer to the user. All three companies, Microsoft, Google and Apple are pushing forward with AI assistance, each in their own way. The shortcomings of Allo should not distract investors from the fact that Google has a significant lead over its competitors. More than anything else, Allo was intended to build on that lead.
AI is probably the next battleground in personal computing/mobile devices. Here, I'm combining the categories because they really are becoming merged. Although Allo is not a revenue generator per se, Google is moving to insert Google Assistant in basic services such as messaging. We can expect GA to become ubiquitous in Google-provided apps.
This is about making the Android ecosystem "stickier" by providing services that consumers don't yet realize they need. To the extent that the strategy succeeds, it will enable Android to continue to be a growth engine for Google. I rate Google a buy.
Disclosure: I am/we are long AAPL, QCOM.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.