|
Post by icemaster on Feb 18, 2020 13:10:56 GMT -5
I had a very interesting conversation with a very close friend of mine who is in the artificial intelligence industry. I asked the question what is it that a person can do, particularly a judge, that artificial intelligence cannot? His one word response was context.
He said that artificial intelligence doesn't have the ability to handle nuance so those cases that require a nuanced opinion based on the overall context are going to be the one the judges will have to make decisions.
So his theory was that would probably will happen is that there will be another layer between the last denial and when the case comes to the judge to screen out those cases that artificial intelligence could make a decision on before they reached the judge which ultimately means fewer cases that judges will have to hear.
Which invariably may mean a much smaller workforce of judges.
Again, just a theory
|
|
|
Post by nylawyer on Feb 18, 2020 17:04:50 GMT -5
Here's the thing- no matter how sophisticated AI gets, it will never be able to be better than the inputs it receives. So, as long as claimants and reps don't bother to make sure all of the medical records are in the file until right before the hearing with the ALJ is held, AI can't really do much to help
With regards to the Johns Hopkins people- they came to my office to exam things, but they didn't talk to me and as far as I know they didn't speak with any other ALJs either (in fairness, we were likely all either teleworking or in hearings that day).
But I'd be happy to share my thoughts with the consultants- and, in addition to my experience as an ALJ in the agency, I happen to be both a Johns Hopkins graduate (Go Blue Jays!!) and hold an MBA myself.
|
|
|
Post by carrickfergus on Feb 18, 2020 18:52:08 GMT -5
I think "HAL" (Hearing Algorithm Laborer) should have to act as a legal assistant and then a writer, and get exemplary performance evaluations, before being allowed to exercise any decision-making authority.
|
|
|
Post by nylawyer on Feb 18, 2020 18:56:11 GMT -5
I think "HAL" (Hearing Algorithm Laborer) should have to act as a legal assistant and then a writer, and get exemplary performance evaluations, before being allowed to exercise any decision-making authority. ALJ- "I find the claimant to be disabled" HAL- "I can't let you do that Dave"
|
|
|
Post by carrickfergus on Feb 18, 2020 19:19:41 GMT -5
|
|
|
Post by icemaster on Feb 19, 2020 6:58:40 GMT -5
Here's the thing- no matter how sophisticated AI gets, it will never be able to be better than the inputs it receives. So, as long as claimants and reps don't bother to make sure all of the medical records are in the file until right before the hearing with the ALJ is held, AI can't really do much to help With regards to the Johns Hopkins people- they came to my office to exam things, but they didn't talk to me and as far as I know they didn't speak with any other ALJs either (in fairness, we were likely all either teleworking or in hearings that day). But I'd be happy to share my thoughts with the consultants- and, in addition to my experience as an ALJ in the agency, I happen to be both a Johns Hopkins graduate (Go Blue Jays!!) and hold an MBA myself. Great points. Your timing is wild because at the same you posted this, I was having the exact conversation with my friend about the inputs and data (which was one of his points) relating to Artificial Intelligence. The interesting thing he and I discussed was that he didn't believe that our jobs would go away (even though he did agree with a significant reduction in the number of hearings) but he did say there will be a fundamental shift in terms of WHAT we actually do. I'm still young enough (and I'm holding on to my youth, lol) to have witnessed the agency go from paper to paperless, from hard copy to CD to electronic access, and from judges traveling to locations to judges appearing via video. All within a 15 year window. I say that because the culture is rapidly changing and technology is playing a huge role (I.e. CSU, Virtual Detail, Time / Place of hearings). My concern is not so much about the elimination of our jobs in the efforts to increase efficiency whether it be through AI, improved processes, or technological advancements (even though it is a concern) but ensuring we have integral part in communicating the necessity for our input in contributing to increasing efficiency and identifying those areas where nuance, exception and context requires human intervention which can not be accounted by technology, especially AI (no matter how much data the system has).
|
|
|
Post by arkstfan on Feb 19, 2020 9:37:23 GMT -5
I think AI could replace the decision making element at DDS.
It could replace it at the ALJ level you just will need some human layer before you get to a level where the next appeal is a district judge (or more realistically a magistrate). I would expect wholesale remands if AI was last stop before judicial appeal.
|
|
|
Post by statman on Feb 19, 2020 10:21:46 GMT -5
Will not happen, Lie detector technology has been around for years and is not admissible in any court. AI cannot pass a Daubert test even as evidence. Also it will be too easy to game the system once the algorithm is known.
|
|
|
Post by roymcavoy on Feb 19, 2020 10:29:44 GMT -5
Will not happen, Lie detector technology has been around for years and is not admissible in any court. AI cannot pass a Daubert test even as evidence. Also it will be too easy to game the system once the algorithm is known. not only that, but it would take congress to amend the CFR to remove the requirements for an ALJ to hear cases. Doing that with the current Dem controlled house in light of due process considerations is unlikely. Which leaves adding another layer to the disability process — anyone else see any irony in complaining about AVG PROC TIME for umpteen years, getting it down to close to the level you have set, and then adding another layer that would cause it to increase...? Short term, I think the most reasonably expected use of such a tool would be to replace either the initial or recon level hearing.
|
|
|
Post by lurkerbelow on Feb 19, 2020 10:30:34 GMT -5
I'm a tech person, and I love technology...but I don't think an injection of computer programming into the decision-making process is going to help, at all. Technology moves at a ruthlessly fast clip. People generally do not appreciate being treated ruthlessly, especially in a disability decision. I might also add that securing such a system in SSA's systems would probably be a nightmare given what it would have to access.
And the implications of a hacked AI decision-maker should be brutally obvious to anyone. Also, do AIs need to have precedent set for them by the Agency? Will it be done in a public manner? I'm betting not.
|
|
|
Post by icemaster on Feb 19, 2020 10:40:57 GMT -5
I'm a tech person, and I love technology...but I don't think an injection of computer programming into the decision-making process is going to help, at all. Technology moves at a ruthlessly fast clip. People generally do not appreciate being treated ruthlessly, especially in a disability decision. I might also add that securing such a system in SSA's systems would probably be a nightmare given what it would have to access. And the implications of a hacked AI decision-maker should be brutally obvious to anyone. Also, do AIs need to have precedent set for them by the Agency? Will it be done in a public manner? I'm betting not. [b And this is why it is so important to have people like who have skill sets before they were judges as a part of the information gathering and input process. The danger comes when you don't have individuals with specialized experiences like yourself who can extrapolate how technology could potentially be more harmful than helpful and how the need for independent decision making in certain circumstances is crucial.
|
|
|
Post by barkley on Feb 19, 2020 12:50:57 GMT -5
Would it not be ironic if John Hopkins actually did a real work analysis and concluded that in order to increase accuracy, ALJs would need to conduct in person hearings and limit production to 30 cases a month in order to have proper hearing preparation and time to fully draft instructions and edit cases? Would it not be ironic if they determined that decision writers needed time to evaluate cases and work at 80% of current guidelines?
Both would dramatically increase accuracy.
|
|
|
Post by nylawyer on Feb 19, 2020 13:51:14 GMT -5
A little too ironic
|
|
|
Post by statman on Feb 19, 2020 14:14:03 GMT -5
A change in requirements will reduce the number of persons receivingv benefits assuming the eligibility requirements are made more restrictive. AI may reduce the number of persons or increase it. There is no way of knowing but, based on the law of unintended consequences, I would bet it is pro claimant.
|
|
|
Post by arkstfan on Feb 19, 2020 15:27:37 GMT -5
A change in requirements will reduce the number of persons receivingv benefits assuming the eligibility requirements are made more restrictive. AI may reduce the number of persons or increase it. There is no way of knowing but, based on the law of unintended consequences, I would bet it is pro claimant. Initially I suspect it would be heavy on the denials and with remands would come programming changes... then we get your predicted result.
|
|
|
Post by nappyloxs on Feb 20, 2020 23:34:15 GMT -5
AI has been discussed for years, well before Insight. When I heard about it years ago, AI would be used as DDS to help make quicker and more accurate decisions. Hitmer would be expand so a program could analysis the records and assist with the determination. Thus, there would always be an appeals level. Insight is data mining as part of the process. One use of Insight is gathering information on combinations of impairments with a high rate of awards which is used for DDS and policy. Another use is gather information on typical RFCs for impairments again for use in policy making and DDS. When I first heard about AI and DDS, I think they said they hoped to have such a system in place by 2025, maybe later as I heard this years ago. This information was from very high executive of SSA and has been in the works well before I heard it. I want to say it has been an Agency goal prior to 2010.
As to John Hopkins study, this has been in place for months, possibly a year now. They are reviewing every aspect of the hearing process and providing “objective” feedback. If I recall correctly, they are reviewing not just process, but structure, spending, hearing rooms, etc. I am not sure what other parts of SSA they are reviewing. The team was working regularly with select hearing offices, but I haven’t checked on it for months. From what I heard, based on the Navy study, the Navy’s backlog decreased tremendously and was a great success.
|
|
|
Post by pumpkin on Feb 20, 2020 23:56:40 GMT -5
Why doesn’t someone ask the citizens of the Great Lakes State what they think of automating government benefits programs? “Between 2013 and 2015, a disastrous automated computer system used by then-Gov. Rick Snyder's administration wrongly accused thousands of people of collecting excessive benefits based on discrepancies in reported earnings, hours worked and other information. Although $21 million was refunded to those who were forced to repay money and substantial penalties, the state is the target of lawsuits claiming due-process rights were violated while some tried to untangle themselves from the mess.” www.usnews.com/news/politics/articles/2019-12-05/michigan-victims-wrongly-accused-of-fraud-win-again-in-court
|
|
|
Post by nappyloxs on Feb 21, 2020 14:57:50 GMT -5
waysandmeans.house.gov/media-center/press-releases/house-and-senate-committee-leaders-call-ssa-abandon-proposed-rule HOUSE AND SENATE COMMITTEE LEADERS CALL FOR SSA TO ABANDON PROPOSED RULE CHANGING THE AGENCY’S APPEALS PROCESS Feb 19, 2020 Press Release WASHINGTON, DC – Top Democrats from the House Ways and Means Committee, House Committee on Oversight and Reform, House Judiciary Committee, and Senate Finance Committee yesterday called for the Trump Administration to withdraw a Social Security Administration (SSA) proposal that would make it more difficult for eligible Americans to receive their Social Security benefits. The proposed changes to the SSA appeals hearing process would compromise claimants’ and beneficiaries’ due process and potentially limit their access to their earned benefits, put unqualified judges in control of deciding appeals cases, and contradict the congressional intent of the law governing such proceedings. “SSA’s proposed rule would erode due process for Americans who are appealing a denial of Social Security or Supplemental Security Income (SSI), threatening access by eligible individuals to disability, retirement, and survivors’ benefits,” the members wrote in their comment letter. “Replacing independent Administrative Law Judges (ALJs) with Administrative Appeals Judges (AAJs) is contrary to congressional intent for impartial SSA hearings, and it is not supported by the rationale asserted in the proposed rule.” The letter was signed by House Ways and Means Committee Chairman Richard E. Neal (D-MA), House Committee on Oversight and Reform Chairwoman Carolyn B. Maloney (D-NY), House Judiciary Committee Chairman Jerrold Nadler (D-NY), House Ways and Means Social Security Subcommittee Chairman John B. Larson (D-CT), House Ways and Means Worker & Family Support Subcommittee Chairman Danny K. Davis (D-IL), House Judiciary Subcommittee on Antitrust, Commercial and Administrative Law Chairman David N. Cicilline (D-RI), Senate Finance Committee Ranking Member Ron Wyden (D-OR), and Senate Finance Social Security Subcommittee Ranking Member Sherrod Brown (D-OH). The full letter can be found HERE.
|
|
|
Post by sunshinedaydream on Feb 24, 2020 11:43:48 GMT -5
Monday, February 24, 2020
www.ssa.gov/news/press/releases/2020/#2-2020-1
Social Security Modernizing its Disability Program Decades Old Rule Updated to Reflect Today’s Workforce
Social Security Commissioner Andrew Saul announced a new final rule today, modernizing an agency disability rule that was introduced in 1978 and has remained unchanged. The new regulation, “Removing the Inability to Communicate in English as an Education Category,” updates a disability rule that was more than 40 years old and did not reflect work in the modern economy. This final rule has been in the works for a number of years and updates an antiquated policy that makes the inability to communicate in English a factor in awarding disability benefits.
“It is important that we have an up-to-date disability program,” Commissioner Saul said. “The workforce and work opportunities have changed and outdated regulations need to be revised to reflect today’s world.”
A successful disability system must evolve and support the right decision as early in the process as possible. Social Security’s disability rules must continue to reflect current medicine and the evolution of work.
Social Security is required to consider education to determine if someone’s medical condition prevents work, but research shows the inability to communicate in English is no longer a good measure of educational attainment or the ability to engage in work. This rule is another important step in the agency’s efforts to modernize its disability programs.
In 2015, Social Security’s Inspector General recommended that the agency evaluate the appropriateness of this policy. Social Security owes it to the American public to ensure that its disability programs continue to reflect the realities of the modern workplace. This rule also supports the Administration’s longstanding focus of recognizing that individuals with disabilities can remain in the workforce.
The rule will be effective on April 27, 2020.
|
|
|
Post by statman on Feb 24, 2020 21:42:01 GMT -5
Do you still include inability to communicate in English when your give a hypothetical to a vocational expert. if so, there is less here than meets the eye.
|
|