Microsoft secured a patent chatbot that will enable us to talk with dead people

Bruchou & Funes de Rioja - June 13, 2021
Intellectual Property, Privacy, New Technologies and Legal Advertising

Will we soon be able to talk to the dead? Or with a clone of ourselves? There’s a good chance that we will. In December 2020, Microsoft Technology Licensing LLC secured a patent in the United States that protects the process to create an artificial intelligence able to simulate the personality and way of speaking of any human being. Soon, this technology will enable us to converse with digital clones of historical figures, loved ones who are no longer with us, ex-partners we have split up from, our nation’s politicians and public figures, and ultimately with copies of any person we are interested in talking to. Possibly, as appears from the patent, users will be able to create clones or generate a new version of themselves at an earlier stage of life in such a way that, for example, a fifty-year-old businessman will be able to converse with himself, but at twenty, when he was still a student.

The process to create a clone will require the user’s input. Drawing on a huge variety of information categories that include pictures, voice audios, social media posts, e-mails, letters, transactional information, geo-tracking, gender, age, preferences, opinions, interests, profession and other behavioral data, the system will generate a digital replica that will not only be able to write like the copied person, but appear on screen with a 2D or 3D interface, that is, with a “face”, and even talk with a voice that resembles the original one.

Indeed, artificial intelligence such as this will lack consciousness, or, at least, consciousness as cognition of one’s existence. However, it will certainly be able to pass the Turing Test in its most demanding variants, like so many other previous artificial intelligences. That means that, under certain circumstances, it will be able to dupe other human beings into believing it is not a machine but a flesh and blood creature, an equal.

Against this kind of technological development, people usually have conflicting reactions: on the one hand, some show real enthusiasm; on the other, there are voices of caution. In this case, the resemblance of the patent’s process description to works of fiction in which attempts to create artificial life lead to horror, from Mary Shelley’s Frankenstein to a recent chapter of the Black Mirror series, no doubt contributes to publicize Microsoft’s invention, but also generates uncertainty and gives rise to a proliferation of staunch criticism.

From a strictly legal point of view, it is debatable whether software such as the one described in Microsoft’s patent is one hundred percent compatible with the right to privacy and personal data protection. Both in Argentina as in many countries of the world, there is fairly robust legislation that curtails possible uses of our information, even what we post on social media. In the case of a deceased persons’ data, although laws regarding data protection are ambiguous, neither are they null. At least, data processing should require a legal authorization, such as the consent of the person who is to be «cloned», or any other legitimate and adequately substantiated interest. Likewise, there should be a notification mechanism to inform people who have been potentially «cloned» about the right they have to access and remove digital records.

Nevertheless, and even assuming a privacy impact assessment, this may not be enough to justify software commercialization that allows anyone to meddle with someone’s data in potentially intrusive ways, without a real possibility of enforcement. But in our country, as in many others, there is no specific legislation relating to artificial intelligence. Even if many countries’ legal systems already include the principle of privacy by design, the majority of data protection standards do not apply to software design in itself, but to its implementations. This also implies restrictive boundaries to the regulator’s intervention.

Notwithstanding the deeper analysis any artificial intelligence developer should carry out regarding the potential impact on each person’s dignity and rights, today Argentine legislation lacks effective mechanisms to guarantee respect of those rights, nor to guide developers when making decisions. Lastly, we must not forget that international circulation of software solutions through the Internet makes it increasingly necessary to agree on supranational and common legal frames.

In the final analysis, in a world of intelligent artificial beings, intelligent legislation will also be required, that is, laws that promote artificial intelligence development, but also know how to protect people’s fundamental rights. It’s not about denying innovation, but that innovation serves society’s interests. In that sense, it’s worth asking: Do we really want to talk to dead people or to a clone of ourselves?



Pardo, Dámaso
Partner Intellectual Property, Privacy, New Technologies and Legal Advertising.

Otero, Juan Agustín
Associate Intellectual Property, Privacy, New Technologies and Legal Advertising.